The FP&A Guy

View Original

The Role of AI in Modern Finance

See this content in the original post

Show Notes

TLDR: Jim O'Neill, co-founder of SaaSWorks, discusses the critical role of AI in modern finance, emphasizing the importance of clean data, the convergence of AI with other technologies, and the need for businesses to adapt to an AI-driven landscape.

These show notes were written by Claude.AI Sonnet LLM Model. 

Dive into the Future of Finance with AI Pioneer Jim O'Neill

Are you ready to revolutionize your business with AI? Join us as we explore the cutting edge of financial technology with Jim O'Neill, co-founder of SaaSWorks and former HubSpot executive. In this eye-opening episode, Jim unveils the secrets to harnessing AI's power in finance and beyond.

Key Insights:

  1. The make-or-break factor for AI success: pristine data

  2. AI's game-changing fusion with quantum computing, nanotech, and gene editing

  3. Why your company needs an AI budget – now

  4. Navigating the ethical minefield of AI advancement

  5. Cultivating an AI-driven culture of innovation

Don't miss Jim's invaluable wisdom on preparing your business for the AI revolution. Learn how to stay ahead in a rapidly evolving technological landscape and transform your financial operations.

Follow Jim:

LinkedIn: https://www.linkedin.com/in/jim-oneill/
Website: https://www.saasworks.com/

Join hosts Glenn and Paul as they unravel the complexities of AI in finance. 

Follow them on LinkedIn:

Glenn: https://www.linkedin.com/in/gbhopperiii 

Paul: https://www.linkedin.com/in/thefpandaguy

Sponsored by QFlow.ai – Revolutionizing B2B revenue planning and analysis. Learn more at Qflow.ai/future-finance.

In honor of having the show notes written by AI we have added the below sonnet as Claude.AI, Sonnet model wrote the notes. 

Sonnet:

In finance's realm, where numbers reign supreme, A new force rises, born of silicon, Artificial minds, a waking dream, O'Neill speaks of change that's just begun.

Clean data's key, the lifeblood of AI, As quantum leaps with nano-innovations dance, In gene-spliced futures, new worlds we'll spy, While ethics guard our technological advance.

Businesses must adapt or fade away, Embracing risk and AI's transformative might, O'Neill's wisdom lights the coming day, Where human and machine in harmony unite.

So heed this call, let Future Finance guide, To prosperous tides on which we'll proudly ride.

In today's episode:

[00:52] - Episode Overview and Sponsorship

[01:38] - Monologue

[07:31] - Your Job is Safe

[08:56] - Importance of clean data for effective AI implementation

[09:58] - Guest Introduction

[15:30] - Founding SaaSWorks and its mission

[18:11] - Common issues with spreadsheets and data management

[21:09] - Discussion on generative AI in business operations

[25:55] - Using AI for innovative and unexpected solutions

[29:11] - Recommendations for companies adopting AI

[32:07] - Discussion on AI adoption and Excel parallels

[32:37] - Random AI-Generated get to know you questions

[36:08] - Conclusion

Full Show Transcript:

Host: Glenn Hopper:: We're practically jacked into the matrix, bouncing around the digital globe with the entirety of all human knowledge at our fingertips.

 

Host: Paul Barnhurst:: Many companies' data is a mess, and if your data is a mess, AI isn't going to work very well.

 

Guest: Jim O’Neill:: If you just let a GPT do coding for you, that should become part of your toolkit. The actual code. Because if you have it, do the same thing again, you may get a different code base. Writing code is like cooking a general recipe. And you know what you're making. You're making paella, you're making a steak or whatever, and you kind of know what you're going to get. You have some variations.

 

Host: Paul Barnhurst:: Can you talk about what it was like?

 

Guest: Jim O’Neill:: So I went to the next management meeting and I said, I need three people, a half million dollars of payroll, and we can't go public until I get this. And I very rarely demanded anything.

 

Host: Paul Barnhurst:: How do you stay creative and innovative in your work?

 

Guest: Jim O’Neill:: So I'm extremely paranoid, and I always think something's going to work harder or be smarter than me.

 

Host: Paul Barnhurst:: Welcome to the Future Finance show, where we talk about.

 

Host: Paul Robotics:: Why Rogers and Tony wouldn't let an expense rocket fuel.

 

Host: Paul Barnhurst:: Future Finance is brought to you by QFlow.ai, the strategic finance platform, solving the toughest part of planning and analysis, B2B revenue. Align sales, marketing, and finance seamlessly, speed up decision-making, and lock in accountability with QFlow.ai.

 

Host: Glenn Hopper:: I'm not typically one for hyperbole or grandiose declarations, but can we pause and take a beat? The buzz for the last couple of years has been all about generative AI and the race toward artificial general intelligence, and beyond, but other 21st-century tech advancements could be just as revolutionary. We've normalized these breakthroughs as business as usual, but today's norm is light years beyond where we were even at the beginning of the millennium. Remember when we marveled at flip phones and email? Now we're practically jacked into the matrix, bouncing around the digital globe with the entirety of all human knowledge at our fingertips. We've progressed from smartphones to smart homes, self-driving cars, Mars missions, reusable rockets, CRISPR gene editing, quantum computing advances, expanding research on nanotech, cancer treatments, and even human trials for brain-computer interfaces. We're approaching a transformation that could rival the discovery of electricity, our realization of our place in the cosmos. This technological shift could make the Enlightenment seem like a flash in the pan. A breakthrough in any of these areas individually could dramatically alter humanity's course. We might be closer than we think to curing diseases, slowing aging, colonizing other worlds, and unraveling universal mysteries. Many of these possibilities sound like science fiction, but they're increasingly within our grasp. Consider the tech stream's converging AI reaching human-level intellect within years, according to some estimates.

 

Host: Glenn Hopper:: Quantum computing could revolutionize fields like drug discovery and cryptography. CRISPR opens doors to treating genetic diseases and even resurrecting extinct species. SpaceX is deploying reusable rockets and is seriously focused on the colonization of Mars. Neuralink's human trials might redefine human-machine interaction, and nanotechnology manipulates matter at the nanoscale for things like targeted drug delivery and ultra-efficient solar cells. Each field is revolutionary alone, but the real Mindbender is what happens when these technologies start to overlap and intertwine. It's not just about parallel tracks of innovation, it's about the explosive potential at the intersections. Picture AI algorithms are optimizing genetic editing, quantum computers simulating complex biological systems, and nanotech sensors feeding data to brain-computer interfaces. This isn't some far-off sci-fi dream. It's the convergence we're barreling towards. These technologies aren't developing in isolation. They're intertwining, creating a technological perfect storm that could redefine, well, everything. We're not just adding these technologies together, we're multiplying their effects. AI-powered quantum computing could crack problems that we can't even conceive of today. Nanotech and gene editing could team up to create superhuman abilities. Space exploration tech combined with AI could accelerate our understanding of the universe exponentially. It's like we've been playing with building blocks, and suddenly we've unlocked the Lego Millennium Falcon set.

 

Host: Glenn Hopper:: The possibilities are wild. The synergies between these technologies could create a chain reaction of innovation, each breakthrough amplifying and accelerating the others. It's not just about what each technology can do on its own, but how they could work together to create something greater than the sum of their parts. But we're not just observers in this process. We're the architects. The choices we make now will shape the trajectory of this convergence. Will we use it to solve our greatest challenges, or will we stumble into unforeseen consequences? The potential is staggering, but so is the responsibility. Buckle up, folks. The future is coming at us fast and it's going to be a hell of a ride. As incredible as these advancements are, each comes with its own set of risks and challenges. AI could surpass human control, leading to unforeseen consequences. Quantum computing might shatter our current cybersecurity paradigms, leaving our digital world vulnerable. Gene editing could create unforeseen biological consequences, potentially altering the course of human evolution and brain-computer interfaces. Well, the phrase hacking your mind takes on a whole new terrifying meaning, doesn't it? We're playing with Promethean Fire here. The power to reshape our world also comes with the power to inadvertently destroy it. And it's hard now to see who's making decisions around questions like, who gets access to these technologies. How do we prevent the creation of a techno elite, widening the gap between the haves and have-nots? What about privacy? When our very thoughts could be digitized and potentially monetized? How do we ensure that AI remains aligned with human values when it might eventually surpass human intelligence? I don't know how far off any of these technologies are, but it seems like now is the time to start considering the answers to these questions before the genie is out of the bottle and we're knee-deep in an episode of Black Mirror.

 

Host: Glenn Hopper:: It's not just about what we can do, it's about what we should do. We need ethicists, policymakers, and the public at large involved in these discussions. Not just the techno bros and massive corporations that are funding this research. How and when are we going to put in place things like robust governance structures and regulatory frameworks to keep guardrails around this? I'm not talking about slowing progress, but I do like to imagine this tech rocket has some kind of steering mechanism We need clear milestones and checkpoints. We need to set boundaries and stick to them. And yes, we need to be ready to pull the plug if things go sideways.

 

Host: Paul Barnhurst:: Welcome to your job is safe for now. This week, I want to talk about a situation where your job is safe for now, but that's not a good thing. Many companies' data is a mess and if your data is a mess, AI isn't going to work very well. AI needs good data to get answers. Your company data needs to be organized. It needs to be clean, it needs to be structured. And the reality is many companies' data is still a mess. Ask the average FP&A professional and the majority of their time is spent on non-value. Add activities I've seen from 40 to 70%, of which the majority of that are, data cleansing, data structure, and trying to get the data in order so you can do the value-added analysis and gain the insights. That's not going to work with AI. AI is going to need good data so it can give you the right answers. How long have we been talking about digital transformations? What, 30 years now? I mean, it feels like it's been a buzzword as long as I've been in FP&A. And the reality is, if your company has not moved into the new millennium, it's not cleaned up your data. You're about to get left behind. Sure, your job may be safe, but your company may not be, so that's not necessarily a good thing. In other weeks, we've talked about some of the things AI has done that you're like, wow, it did that.

 

Host: Paul Barnhurst:: That's bad here. We're going to talk a little bit differently about how your job is safe because you can't implement AI. That's a bad thing. The reality is companies need good data. You need that data model, the data dictionary, and your master data. You needed to be structured that one source of truth, right? The Holy Grail, one ring to rule them all. So if you're struggling, now's the time. Don't think. Oh, as long as our date is a mess, I'm safe. AI won't take over. The reality is, you want me to be more efficient, to be more productive. So start today. You haven't done the transformation. Make the transformation happen even if your job is safe for now. Thanks for joining me for this week's section of your Job is safe for now. All right. I'm here with Glenn Hopper, my co-host, and we're super excited for this week's episode of Future Finance. We have here with us Jim O'Neill. Jim, welcome to the show. Thanks for having me. Very excited to have you. And so, Jim, we want to start off by jumping into some questions. And the first question we'll ask is just can you tell us a little bit about yourself? Give us a little bit of your background.

 

Guest: Jim O’Neill:: Yes. So a former coder, but I lost my coding license a long time ago. Kind of stumbled into a startup in the early 90s. Got to ride that startup through a modest acquisition to a large publicly traded company that had hundreds of other software companies and just got very lucky to get a back seat at a hypergrowth company when I think it was pre-internet. So that was awesome. From there, kind of met a couple of founders that were in that same ecosystem. They had built an amazing product and needed to kind of scale the revenue, so got to join them. And then the irony is, we sold it back to that publicly traded company that I left. So I kind of did a rebound with them, but always stayed very close to data security and tech and either a CTO-ish role or a CIO kind of role. And then post facto in that business that we sold. One of those founders happened to be the founder at HubSpot. So I joined them kind of day zero, customer zero through an IPO, spent about 11 years doing everything from CIO to CPO, being people person to CTO, unofficial CSO. So it was a whirlwind of a ride and it was amazing. And then most recently just started my own company because I wanted to take the front seat versus the back seat and just we'll focus on that today talking about FP&A data, cleanliness, and AI, which will be fun.

 

Host: Paul Barnhurst:: You wanted to help drive the car instead of sitting in the back and hoping that the driver would get you there.

 

Guest: Jim O’Neill:: Yes, I want to see if I could actually drive.

 

Host: Paul Barnhurst:: I know I can't, but that's another story. Glen and I will take turns asking questions. We kind of wrestle over who gets how many questions, but we usually settle on three and three. I usually win if it's more than that, we won't talk about that. All right. So first question here. Glenn's laughing. I'm trying to make him laugh. So can you talk about what it was like being part of that founding team at HubSpot? Being in the back seat of the car for a ride from a startup to a global company with a lot over 50,000 customers. When you left, I think. Yeah.

 

Guest: Jim O’Neill:: Yes, exactly.

 

Host: Paul Barnhurst:: So talk a little bit more about that. We'd love to learn what that experience was like for you.

 

Guest: Jim O’Neill:: Yes, it was tremendous. And I think everyone romanticizes and reframes history or rewrites history. It wasn't easy, but it was an amazing group of people who were committed to building something durable. And I don't think we hear about that a lot in the startup community, at least in the current markets. Right. And so the founders, when they started the company said, we want to build a pillar company in Boston. We want it to have a legacy. It was inspired a lot by if folks have heard of the PayPal Mafia, which was PayPal has had a lot of well-known people, some controversial, maybe you could say Peter Thiel is controversial or Elon Musk is controversial, etc., etc. but an amazing group of folks. So the ride there was to build something durable and do it with the customer first. But I think the most interesting lesson that I would reflect on is while most companies want to grow very ambitiously, the founders said we need to grow and think of 10x. So it wasn't the traditional, hey, we're going to double this year. Every year they said, we need to build things for ten times our size. And I think that is why, honestly, they were able to grow so fast because when you think ten x or even bigger inefficiencies don't get optimized, they get radicalized, right? You just cut systems out or you just retool things. And so that's the thing. I'm most grateful for not just talking about scale but thinking of engineering problems. And when we talk finance today, even their FP&A team was a data animal. And because their team was built on the fact that they were going to have to handle ten times the revenue in the next 2 to 3 years. So it was a fascinating experience.

 

Host: Paul Barnhurst:: It sounds like one follow-up question I want to share is when you mentioned PayPal. Kind of a fun story. I worked for American Express, and in the early days, PayPal approached American Express when they were just getting started and said, we want to sell to you guys for $1 million. And American Express said we're not interested. And I look back now and I think American Express and PayPal would have been the perfect combination kind of hindsight 2020. But anyway, you reminded me of that because that was a story that was talked about quite a bit at American Express, of one of those, that's awesome. You wish you had done this type of thing.

 

Guest: Jim O’Neill:: Yes. I guess the other thing on the experience is hiring athletes, and I know people talk about that a lot, but I don't know if people execute it. So I was joking about my own titles at the start. I probably had ten jobs in ten years, and that's pretty hard. And I think that's hard for most people because you become an expert or you become an engineer. I think what HubSpot did a remarkable job was, if you had the appetite and the kind of courage to risk your job because that's what you're doing if you're taking such a different role. The support system was pretty tremendous. And so I think kind of the other cultural thing that I'd love to see more startups do is and especially I think is going to unlock this a lot as we talk about it, is to take bigger risks with fewer safety nets, but more support. And what I mean is, my safety net is like, you are out on the ledge and you've taken the leap of faith, but the support now is you've got experts everywhere. As I look at it is an expert, right? It's unbiased towards you. It doesn't matter who you are, at least today. And you can ask questions in a safe way and get interesting feedback. So I think HubSpot did a tremendous job at hiring athletes, rotating them, and truly giving them a support system to kind of rethink how traditional functions are done.

 

Host: Glenn Hopper:: That's great. So, Paul, I'm going to sneak an extra question in here. I want to move from from HubSpot. And the great success story that was to what you're doing now with SaaSWorks and the nature of which you guys are doing and what you're trying to bring to the market.

 

Guest: Jim O’Neill:: So in the spirit of most great ideas happen on someone else's watch. And I and my co-founder have kind of two versions of this story, but my version is the idea for SaaSWorks, and he just came from an investor angle. I came from an operator angle. I started inside HubSpot. So this is a pretty true story. I don't share too much, but I believe it to be true. 2011 we were getting ready for an IPO or we were contemplating it. And if we all remember, 2008 and 2009 were some tricky years. Markets were still thawing. You know, it was kind of the long thaw. And we talked about what it would take to get ready for said IPO. And I recall a management meeting with the founders and kind of the founding team. And we were talking through numbers. Right. And so you both, as FP&A and experts, like what's your customer count? Pretty simple question. What's my gross dollar retention? What's my net dollar retention? And so as we went through these numbers, my recollection is we had four different variations on that theme depending on if it was the head of strategy and data, the CRO. I distinctly remember leaving that meeting feeling pretty uncomfortable at best, because I knew at the end of the day I was responsible for getting the data right because I was running it and all the infrastructure. And so, Glenn, the reason why this kind of seeds was SaaSWorks is doing and I'll share.

 

Guest: Jim O’Neill:: That was, I knew to go public, we couldn't be wrong. Right. Because that's how, you know people get in trouble. I don't look good in stripes would be the running joke because we're liable for for accurate data and accurate numbers. So I went to the next management meeting and I said, I need three people, a half million dollars of payroll, and we can't go public until I get this. And I very rarely demanded anything. I'm a pretty passive guy. And so the fun part was founders like, you're lying down on the railroad tracks on this one. And I'm like, I don't think we have a choice. And what's fascinating, Glenn and Paul are today 12-ish years later, and the state of the state is still those similar stories of there is no system out there that is your single source of financial and or revenue truth because every business's revenue is slightly nuanced. So we'll give credit to the company. Snowflake I think it was a genius name. Everyone's data is slightly different. And every FP&A professional has to look at their models slightly differently. So there are industry standards for RR and MRR and LTV and KAC, but everyone's kind of rolled up their own of how they treat it. Do I take demos and trials and include them? Do I do paused accounts? Do I have a ramp period?

 

Host: Paul Barnhurst:: So I'm going to ask one quick question before Glenn goes here. Jim. So what do you call a broken spreadsheet since you brought up a broken spreadsheet?

 

Guest: Jim O’Neill:: So I would say there are three answers. We love the number three.

 

Host: Paul Barnhurst:: Real quick though, I'm going to tell you mine. A little bit of a joke, right? You call it the new employees' problem.

 

Guest: Jim O’Neill:: And then six months later they're not the new employee anymore. That's the challenge. And then they.

 

Host: Paul Barnhurst:: Passed that broken spreadsheet on to somebody else. Isn't that how it works?

 

Host: Glenn Hopper:: Go ahead. I just kind of had to have a little bit of fun with you.

 

Guest: Jim O’Neill:: There you are. Right on. so for bigger businesses, I would say, businesses that approach 100,000 to 1 million records in the spreadsheet, tend to break the spreadsheet. Right. And so Excel has a physical limit of a million rows. You can do tricks on it, but a million rows. What happens is as companies try to make this customer file or revenue bridge or revenue waterfall, whatever term you use, over time, the combinations of permutations break Excel, meaning it just can't handle the volume. It becomes too slow. Data corruption. So that's the first flavor. The second flavor is broken formulas and macros. Because what they're doing is oh, but for these products I calculate R this way. And for these products, I calculate it this way. And then I'm also going to do a formula on formulas. But I forgot to change a reference. And then the whole model bus. And so we find that people spend incessant time trying to make sure their models are right because they've used a good but not perfect programming language called Excel to kind of build out these models. That's the second breakage, I would say. Third is change management. These spreadsheets are said spreadsheets, even if you're using Google Sheets, because the volume can fit change history and the volatility of data is real.

 

Guest: Jim O’Neill:: So I'll make fun of us in ops. No one ever goes back and changes anything in Salesforce after the fact, right? We never like Salesforce is perfect or NetSuite is perfect. As soon as you change a historical record, the ramifications of that are real for any of us that have to produce audited or audit quality type numbers. So I think the third is the most dangerous or one small data change upstream can have a massive impact on the business, and it can be a wholesale change of someone changing how they structure their NetSuite or their Salesforce. And it's known about, but they don't understand the downstream impacts or something as simple as a sales rep closes a deal on March 31st, they have approval to get the signature. The next day. Everything's legally good, but that change causes a kind of ripple effect. So that's the way I think Excel gets broken from the size of data to the complexity of formulas and macros. And three change management. And all of those can ruin reporting.

 

Host: Glenn Hopper:: I kind of zoned out for the last part, because as soon as you started talking about sort of the cascading exceptions, my brain went into like a PTSD mode, thinking of just trying to solve for that, where you start with this great model and then as those exceptions come and you keep coming.

 

Host: Paul Barnhurst:: I've had a few PTSD from this conversation. Yes, cleaning data and FP&A professional.

 

Guest: Jim O’Neill:: And if you guys see that I mean, those are my three.

 

Host: Glenn Hopper:: Yes. So I know I'm not asking you to hey, I wash your business as a lot of people are doing out there. But you know, with the nature of what you're doing and dealing with the data and seeing where I'm thinking particularly about generative AI, but just AI in general. I know you guys are using algorithms in the work that you do, but maybe this is how you see your company being able to apply it and use it internally or what else you're seeing. I'm wondering, how you were seeing or how you foresee AI being used to enhance kind of predictive analytics in finance. And I'm thinking particularly in terms of forecasting and risk management.

 

Host: Paul Barnhurst:: Ever feel like your go-to-market teams and finance speak different languages? This misalignment is a breeding ground for failure, impairing 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 be revenue. QFlow quickly integrates key data from your go-to-market stack and accounting platform and then handles all the data prep and normalization under the hood, it automatically assembles your go-to-market stats, makes segmented scenario planning a breeze, and closes the planning loop, creates air-tight alignment, improve decision latency, and ensure accountability across the teams.

 

Guest: Jim O’Neill:: Yes, so we are using it in our product and our operations to benefit the product for our customers' benefit, right? So I think I can be pretty intimidating if you don't have the time frankly. Like it's not there's no free lunch. I think it's magical when we're doing some parlor trick ChatGPT question-answer kind of thing, right? And it's pretty exciting. But if you're trying to do something with high precision, I think it's pretty intimidating, frankly. And so a big part of what we've done, and I'll talk about the operational use of it, is to embed it in some of our workflows with our customers, to introduce it to them so that they feel like they're getting kind of baby steps and introduced into it because there's a bunch of security, privacy, I'm sure. I'd love to hear what your experiences are, but there are a lot of blockers to using it in the corporate world today, right? Does it matter the size of the business? If you have any risk and liability, I'll make fun of my old CSO role like the liability police come in right away. It's like you can't use it, right? But we know 75% of all current knowledge workers are using it per the recent study, and yet only 25% of businesses are paying for it.

 

Guest: Jim O’Neill:: So that means 50% is kind of out in the wild, wild West, and we should just accept that. And so our hacks are kind of chunk things up in them up, give ourselves time schedules validate during the time schedule. Now what I didn't say is once we got that final output, it wrote the Python code, which we then can run every time. And I think that the other thing people aren't doing enough with AI is if you just let a GPT do coding for you, that should become part of your toolkit, the actual code, because if you have it, do the same thing again, you may get a different code base, right? So I also think just like working with an employee, just like working with a developer in the IT team or your product team assets get created. You need to start saving those assets. So that's a very long-winded. I try to be pretty descriptive because I think so many people because you talked about eye washing, it's easy to talk about how amazing AI is. I think it's hard to use it at scale and predictably if you don't have a framework to use it in. Yeah.

 

Host: Glenn Hopper:: Talking about saving off the code, it's I've always been a wannabe coder, and I've built a lot of stuff, not production code, but like Jupyter Notebook, Colab, and everything. But I I will pull off code and throw it into a Jupyter notebook. But you know, the security people and the production people are like, you can't run code in this environment kind of thing. But I think that that's happening probably a lot now, sort of the citizen development where people if your company doesn't have a sandbox for you to do stuff like this, and if there's not a clear policy, then you're getting employees who are becoming cyborgs where they're just doing it on their own and they're throwing out. And that could be a real security problem. And I guess the other thing that I latched on to there is and I do a lot of training on using generative AI for finance people, and I love it when I see that aha moment happen with them. If you're thinking of the way you do something in Excel or the way you architect code if you're writing something out, you have to go through and map it all out, architect it, plan it, and then you know, you know what you're looking for. And so people have a tendency at first to interact with generative AI that way, where they're telling it specifically what to do, where one of the strengths of it is when you leave it open and you let it make determinations because it's going to come up with stuff that may not make any sense to you at first, but then they may find some interesting correlation or something, but you're limiting it if you're just telling it to do what you always do. So that's one of the fascinating outputs that comes from generative AI is when you let it come up on its own.

 

Guest: Jim O’Neill:: Hey, man. And so one thing we didn't talk about, I don't think we do this enough, but I do get around to having an anti-persona. So my partner is a formal CFO and an investor asks AI to do the opposite of what you want it to do. So one thing we do is we will take numbers for customers and we'll run it through GPT with the right security and privacy, etc. We'll then say you are a board member at a publicly traded company. Your job is to find issues with the business model. And like like this is free in my mind because whatever you're paying $30, $50 a month for these tools, it's exceptionally good at being the anti persona because to your point, Glenn, it's unfettered. It doesn't know any better. So I don't think people I think people are still using AI in many safe ways. I think a lot of business people need to start using it in the antagonist way and be the alter ego because you're going to get feedback that you wouldn't get otherwise from a teammate, right? Teammates tend to be nice, not kind Gpts can be pretty kind. They can just tell you exactly what the data says.

 

Host: Paul Barnhurst:: I hadn't thought of that. That's a great point. I'm already thinking of some cases. I might run some transcripts through Glen and see, giving us a negative view of how we did on our episode, and see what it tells us.

 

Host: Glenn Hopper:: Don't our listeners already do that?

 

Host: Paul Barnhurst:: The other thing I was going to add to kind of going back to when you talked about how you need to set time and it's a process with AI, I hear a lot when I train people, it's like, well, it feels like I gotta hack through it. Yes, to a certain extent, you do. It's not going to come back with the perfect answer every time. And yeah, I was trying to create some dummy data as an example for some training I'm doing next week, and it took me quite a while to get it to look like what I wanted. I'm like, okay, every product has the exact same gross profit margin. That's not what I need. Oh, there's these categories that do category one, two, three. So there's areas where you have to be specific, but you also want to allow it to figure out how to do it. And so it was definitely a it's been a learning as I use it more and more to create data for training and all kinds of different places. You just got to be patient and sometimes you may not get the answer you want. There's been a few times where I'm like, fine, I'm just going to do this one on my own. It's not I'm not getting there. And I'm sure if I took more time I would get there. It's just sometimes you're like, oh, I could do this quicker myself. It's like Excel. Yeah. Like automate it or do that copy and paste for a few minutes.

 

Guest: Jim O’Neill:: And that's like the training a new employee and this is someone else's quote. But I liked it there. Like if you're a programmer, writing code is like cooking general recipe. And you know what you're making, you're making paella, you're making a steak or whatever, and you kind of know what you're going to get. You have some variations. Working with AI is like teaching a child, right? It's got so much potential, but there's no precursor. So I love what you just said. If we restrict it too much, it's going to give us what we've told it to give us. If you unleash it, you're going to get it's feedback versus advice. You're going to get advice. You're not going to get feedback, right. So it's not going to tell you how it is going to tell you what to do different.

 

Host: Paul Barnhurst:: All righty. Alrighty. So you know, we're coming near the end of our time. I think we just have 1 or 2 more questions, and then we're going to move into our kind of fun section here. You know, you've talked quite a bit about what you're doing and using AI. What would you recommend out there for companies right now as they kind of go on this journey? Anything you would offer for the company itself? Sounds like we've already talked about how 50% of people are using it without their company. So what should companies be doing?

 

Guest: Jim O’Neill:: So I believe this is someone else's term as well. I'm not that clever. I think there's an advent of AI ops that's going to happen, just like DevOps happened when the cloud came around, right? And building up infrastructure and both of you have exposure. I'm sure it was expensive, right? These are specialized people. And so I think if a company is going to use AI unless they're able to just use commercial AI off the shelf and just give it data, which I think very few people can do, I think there's going to be almost a cultural change in companies that we should all start embracing now on AI ops or AI operations and a very simple thing for finance teams. There's zero reason to give a GPT today your customer names. It can just be deduped or anonymized whatever we want to call it. And then I think lastly is culturally, we're heading into the second half of 2024. I would be budgeting for AI investment in any company for 2025 so that it doesn't become a financial trade-off. What we're doing and we're a pretty early-stage company is zero-based budgeting. No new headcount. If we haven't done an AI bake-off. Meaning can we bring in more AI to prevent hiring more people? That's not about reducing people. It's about leveraging and making our team happier because they have less mundane work. Even in a tiny startup, there's still mundane work to do.

 

Host: Paul Barnhurst:: I'm a startup of one. And do you have any mundane work, Glenn? Because I sure do.

 

Host: Glenn Hopper:: Oh, well, we could talk about this all day and it's going down this road thinking about, I mean, it's almost like a whole other episode. And I know beyond what we brought you on to talk about, but now I'm thinking more about sort of deployment and how companies can have their employees using AI. And I think I did hear I think it was a professor, Ethan Mollick said, something that has stuck with me is, in an IT environment when you're putting in a new ERP or CRM or whatever it is, it comes top down. It is this is the system we're using. Go use it. This is how you do it. But with the nature of AI, it's the way that we use it. This isn't going to come from the top down. It's going to be the front-line employees who are finding ways to streamline what they're doing and finding ways to enhance it. And so the idea of getting it out to the entire company and within their guardrails, I know as a security guy, that's very important. But within the guardrails and letting them find ways to use it to get more efficient for the company.

 

Guest: Jim O’Neill:: Yes, they're going to use it anyway. They really are, and it's so accessible now. So I don't know why we don't adopt it.

 

Host: Paul Barnhurst:: Kind of funny enough, isn't that a lot like Excel in the sense of spreadsheets? Top-down in the sense you put them into the company, but they're using them in all kinds of ways in figuring solutions that you never planned from. Rarely do you have someone at the top saying, okay, everybody has to learn VBA or everybody has to do it this way. And so I think you're kind of seeing a similar thing with AI because it's so flexible and there's so many different things you can do with it. I think when it's readily available, it often does come from the front line. It's going to be very different. I think that's a great point.

 

Host: Glenn Hopper:: Well, Paul, should we move to our impersonal personal section where we have robot-generated questions?

 

Host: Paul Barnhurst:: I was thinking I'll start at this time and I'll let Glenn go second. So we've had a ChatGPT come up with a bunch of random questions that we get to ask to get to know about you. And so we give you a little bit of choice here and on mine. I'm going to give you two options. You could pick a number between 1 and 29. Or you can have the random number generator pick the number for you. And that's the question I'll ask. So which would you prefer?

 

Guest: Jim O’Neill:: I'll go with the random generator.

 

Host: Paul Barnhurst:: All right. Let me come over here.

 

Guest: Jim O’Neill:: And we can't talk about AI and not let AI do a little thing for us, right?

 

Host: Paul Barnhurst:: All right. We have the number 22. So here's the question you get. This is a pretty good one I like it. How do you stay creative and innovative in your work?

 

Guest: Jim O’Neill:: So I'm extremely paranoid. And I always think something's going to work harder or be smarter than me. So to counter-event that I have two kids that are entering or leaving high school, entering college, I do the what would they do? And I pay attention to the tools they're using on social media. I'm probably a free-range parent chicken parent, right? I let them do a lot. And because I do think they think so differently Paul and Glenn, I would hire them tomorrow and tell them to build an AI company or a company based on AI. They have no predispositions to work, they don't know how to work. And that's a that's a blessing right now. So for me, look at the youth, find the tools that they're using, and copy that. And then for me say, what would the 50-year-old me do? And what would the 15-year-old me do? And if I can, a B test that I tend to make more modern decisions.

 

Host: Paul Barnhurst:: Great answer.

 

Host: Glenn Hopper:: Love it. All right, so Paul and I take a slightly different approach to this. I'm more fly by the seat of my pants. So we built a GPT to create these, but I tried to mix it up. First, I think four weeks ago we used Gemini to create questions, and then we used this GPT that we built using ChatGPT 4.0. And I am today I have Claude Opus, and I've asked him to generate some questions for a podcast guest. So the randomness here is whatever the AI generates, let's see what it is.

 

Guest: Jim O’Neill:: Now would only be better if I could answer with an agent myself, but I'm not going to do that.

 

Host: Paul Barnhurst:: I'm sure we'll get a guest that will eventually do that.

 

Host: Glenn Hopper:: Or we can have Alexa and Siri just going back and forth for the show.

 

Guest: Jim O’Neill:: That's the future of buying software. I've heard the agents are going to negotiate with each other to race to the bottom.

 

Host: Glenn Hopper:: Well, yes. All right, let's see what Claude says. Okay, I'll take this one. It's a short answer. Are you reading anything interesting right now? Either personal or professional?

 

Guest: Jim O’Neill:: I don't read a lot of books. I read a lot of blogs. And there is a series of daily newsletters from inside.com. It's a shameless plug for them that gives world news, AI, and deep tech. So I read that religiously every day, and then, probably two hours a day on AI podcasts, just like your AI podcast a couple of sessions ago was awesome, by the way, so I read very little long form, a lot of short forms and a lot of podcasts because I just find it to be more refreshing. The Life 3.0 book that came out about ten years ago, because we're talking did change how I think full stop. So if anyone hasn't read any of the deep I books, I highly recommend it because it changes perspective pretty deeply.

 

Host: Glenn Hopper:: Was a life three point. Was that Max Tegmark's book or? Yes, that was a good one.

 

Guest: Jim O’Neill:: Yes, that did change.

 

Host: Paul Barnhurst:: That one, I haven't read that one. So you guys have to give me a book now.

 

Guest: Jim O’Neill:: It's probably almost ten years old and it's pretty creepy. What's happening? That's all I can say.

 

Host: Paul Barnhurst:: Creepy is probably a good note to end the show on, right?

 

Guest: Jim O’Neill:: There we go. That's on-brand for me anyway.

 

Host: Paul Barnhurst:: Yes, we won't go there. Thank you so much for joining us, Jim. We've had a pleasure chatting with you today and just getting your thoughts and what you're doing at SaaSWorks. And kind of a last question, if anyone wants to learn more about you or get in touch with you, what would be the best way for them to do that?

 

Guest: Jim O’Neill:: jim@saasworks.com, saasworks.com. Very simple.

 

Host: Paul Barnhurst:: That works great. Well, thank you so much for joining us, Jim. It's been a real pleasure. And like Glenn said, I bet we could go a couple of hours, but I'm not sure our audience would like that. We probably have work we have to do. So at this point, we'll let you go. But thank you, Jim.

 

Guest: Jim O’Neill:: Thanks, Jim. Thank you guys, and let's get back to building our agents.

 

Host: 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.