Listen to This if You're Using AI to Optimize Your Business | Ep 245

Speaker A: Everyone in their dog is talking about AI right now, what tools to use, how to automate, and how to stay ahead of the competition. And look, I think AI is super important for every business, so let's just be clear on that. But also, there's this conversation that almost nobody is having, and it's the one that matters the most for your business. It's not about which tool to use, it's about whether those tools are actually helping or. Or hurting your business. Today I want to talk about how you should be thinking about AI when it comes to optimizing your company. On Saturday, I had some free time, so I was scrolling through LinkedIn and I came across a post that was intriguing. It was a video posted by a CEO of a business. And essentially he was saying this, I'll paraphrase, at the beginning of the year, he was looking for a cfo. He realized that he had some gaps in his business as it related to finops, to his forecasting, to his KPIs, et cetera. And he's like, I need to hire a cfo. So he started making a list of all the different things he needed from this individual. But then before he posted, he thought, hm, what hm? If I posted this to AI, what kind of prompt would I get in return? So he did that. So he posted this. He typed in the prompt, hit send, it spit out a bunch of things like a dashboard, some KPIs. It even gave him a cash flow forecast. And so in his post he's like, I just saved 300,000 to $400,000 a year by using AI. And if you want my same hack in my same system, just comment down below cfo, and I'll send it to you. And I started looking through the comments and I realized a lot of other CEOs were commenting, CFO, CFO, CFO, cfo. And this guy, just to be clear, he wasn't really selling anything. He's just trying to be helpful to other people in his network, in his community. And so how great is it that we live in a world where people do just that they're trying to provide helpful resources for other entrepreneurs. That's great. And I commend this individual for this. And I don't think he's doing anything malicious. But I think there's a gap in understanding of how AI actually works and how you can truly optimize a business. And the pitfalls aren't always so clear. So let me help you steer clear of them. When it comes to AI, I'm sure you've heard of this, this term before, but there's this term called false positives, which essentially means that AI will return something. So let's say you upload a spreadsheet or you say, look at this code. It'll come back and say, oh, uh, I found a few issues, or here are some things to correct. But then you go in and you dive into the model and you realize everything's okay. That's called a false positive. So it's saying something may be wrong or there's an issue, but there's really no issue. Now, when I did my research and I looked across all the different models from OpenAI to Claude to Grok, did you know that on average. Right. Uh, on average, the most sophisticated and advanced models today return somewhere between 20 and 25% false positives? Now, just as a caveat and a note, that number may be a lot lower depending on how you prompt AI. But that's just the average. So think about that for a minute. Let's just relate that to a doctor's visit. Let's say you go to the doctor, and the doctor that you trust that you've been going to for years, has a track record of false, uh, positives of 20 to 25%. And then you go and you realize something's wrong with your leg, Right? And he's like, we need to amputate this. And so he cuts off your leg. And then after the surgery, he's like, oh, no. I just realized I read your MRI wrong. I thought that was a lesion on your leg that we needed to cut off. But in reality, it's just a tattoo. Sorry about that. Now, I'm being a little overdramatic here, but I want you to understand this point. The same thing is true with Waymo. I don't know if you've seen these driverless cars, but can you imagine if a Waymo. Let's just say they're able to reduce their false positives down from 20 to 25%, down to 10%. I don't know about you, but I would not be putting my kid in a Waymo with that type of track record. In fact, the false positive rate needs to be, like, 0.00 something percent before I feel comfortable putting somebody that I love and I care about in that type of situation. So think about what the CEO, uh, was offering to other people. And perhaps in his business, he doesn't know what he doesn't know. Especially at $20 million. I'm sorry, but I don't think you could just have an AI driven CFO in your company, a virtual cfo, right, Built by an AI agent, telling you everything you need to know. Perhaps you can, but I just don't think we're there yet. Especially when it comes to false positives. All right, so let's play this out. In your business, you need to understand your cash flow. This is a critical thing for most business owners in order to understand free cash flow. And to build a cash flow forecast, let's just take it down to the 13 week level because that's often how I build cash flow forecasts on a 13 week basis. It's very complicated to build a cash flow forecast and here's why. You don't control when your customers pay you. That's why cash flow forecasting is so complicated. I could be the smartest mathematician in the world, have all these, uh, statistical regressionary blah, blah, blah things in my model. But guess what? I don't control when ABC Customer pays me. It's in their control. And if your average order value, your average contract value or whatever it may be that you're selling, if it's big, the bigger the variance is going to be. So it's just imagine you're waiting on a, ah, $200,000 payment from a customer and you need that in order to cover your payroll. But guess what? They forget to submit the payment on time or the checks in the mail or whatever it may be. And you're expecting that because that's what the AI is telling you. That's when it says it's going to land. But there has to be human judgment built in. And I don't know how AI is going to do that, at least right now. Therefore it may return a false positive. So there's so many nuances. Trust me, I have spent thousands and thousands and thousands of hours forecasting and in the financials. And I could tell you there are a lot of nuances. You literally have to download your accounts receivable, aging by invoice, and you have to go line item by line item. Um, if you want an accurate 13 week cash flow and you have to say, when is this company going to pay me? Oh, I talked to Frank the other day. He said that he's going to cut a check in two weeks. Oh, I talked to Betty the other day. She said she's going to pay early. She's actually wiring the payment on Thursday. That's the kind of detail that needs to be had when it comes to building out these forecasts. I don't know how an AI agent is going to do that in your business. And if it does, let's just think about the average false positive rate. Do you want to bet your business on that? You think you're good. You're like, huh? Oh, I'm great. The agent said, um, I'm, I'm positive cash flow throughout the, the end of the year, easy selling. You go on vacation, you start taking distributions, you overspend on Capex, money's trapped in your working capital. But the false positive from your AI agent is telling you you're okay. You don't want to risk your business, you don't want to risk your solvency based on something like this. So that's what's really important for your business. I've also shared other ideas when it comes to KPIs. Some people will listen to this podcast and they'll think, I need to have KPIs in my business. So they'll go to their favorite AI platform. They'll type in, what KPIs should I be tracking for my electrical business, for my plumbing business, for my hair salon, for my restaurant, whatever it may be, and therefore, I'll spit out a list. And you're like, boom, got it. All right, now I'm going to pull the data, calculate it. Now I got a KPI dashboard. I'm all set. But those KPIs, if they don't tie back to your strategy, they're pointless. Like, you can measure whatever you want. Like, I can measure the temperature in this room. I can measure how many steps it takes me to go from the bathroom to the lunchroom. I can track a bunch of random things. But if it doesn't connect back to my central goal, to my strategy, they're pointless. And more importantly, you need to be tracking things where you can change behavior. You need to be tracking things that will answer a question that you have in your business. Too many owners, they have KPIs. But then when I say, what do your KPIs tell you? They can't answer that question. Case in point. I just met with a company not too long ago, and we sat down in a room and I said, pull up your financials and pull up whatever information you share in your monthly financial reviews. Because they said they had a really good routine. They're reviewing financials on a regular basis. And they did. They pulled it up on the screen, and the cfo, uh, was just scrolling line after line after line, going between rows and columns and graphs and charts and all this stuff. And he got done And I said, how do you know what's important? And if this number you're measuring, this budget versus actual, and you have these variances and you communicate that to your general managers, then what? They're just like, ah, yeah, I'm gonna go try harder. Thanks for telling me, like, what's coming of it. So when it comes to AI, there has to be context. There has to be context behind it. So when you're doing a prompt, if you don't have the context and you don't have the knowledge behind what you're looking for, you can just put your financials in there and say, how am I doing from a financial performance standpoint? And it's going to spit out some jargon, right? You put in some KPIs and I'll tell you some trends. But if you're not sharing the right context, you're going to get false positives or you're going to go down a path that's going to waste a tremendous amount of time, energy and capital. So you have to have the knowledge to know how to prompt, you have to have the knowledge to know how to diagnose, and you have to have the knowledge to know how to interpret. Oftentimes talk about the story behind the numbers. Sure, AI could do a lot of wonderful things. And I'm not arguing against AI. I think AI is great. I'm just trying to point out the false positives, the context, and everything else that you need to make it a useful tool in your business. If not right, you're going to be using it, you're going to be getting these numbers, these KPIs. But if you don't take the action, and if you don't take the right actions, what does it serve you? So that's really important. So there's no replacement for knowledge that's really critical. And like I just mentioned, you don't want AI to lead you down a rabbit hole. So this is another thing I've seen with a lot of companies. They'll use AI. AI will spit off some things. They start having a conversation. But remember, you're just having a conversation with an algorithm that's taking ones and zeros and some complex math, and it's returning information that's based on mathematical equations in large language models. It, it doesn't understand the nuances of your business. Think about that. You may say, what do I need to do in order to sell my business? And it responds, these are the five things. And it could give you context, it could build you a plan it could do a bunch of amazing things. But if it doesn't understand that you need to sell the business in this time frame, in this state, with this tax situation, because your son has this unique thing in, your wife has this and this and this, whatever it may be, that doesn't understand all the nuances of your life, then it's going to send you down this really bad path, especially if you're trying to turn around your company. I've had businesses approach me like Steve, we're really struggling with cash flow. We need to turn around our business. And then they start building out these models. They start relying on AI to tell them what to fix. And AI tells them go and cut their overhead. But it's not an overhead problem, it's a pricing problem. And more specifically, it's an offer problem. Their offer is terrible. Customers don't understand it, or it's an estimating problem. But the problem is embedded in the template down on row 42. Because there's this or that, right? That's the specificity that I'm talking about. That's the context that you need so you don't go down a rabbit hole. Otherwise, you're going to waste a lot of money, a lot of time and a lot of attention. Have to get narrowly focused. Which gets into my last point. AI is just a tool. Just like computers were a tool and they replace people who were doing things by hand. So AI is going to replace a lot of people who are using old tools or following old systems, and they're not updating their skill sets and their capabilities, but it's just that it's a tool. It makes a CFO even more powerful if they know how to use it. It makes a business owner more, more effective if they know how to use it. But you can't take somebody who doesn't have the context and doesn't have the knowledge and just put them in a position and say, hey, you're now our, uh, cfo. You have no accounting background, you have no financial background, and you're going to be our CFO and you're going to use OpenAI ChatGPT to do your job. It doesn't work that way. There's so many nuances. You have to understand how everything works and how everything interconnects. So AI is not going to save a business with the wrong fundamentals if margins are wrong, if job costing is off, and if you have no financial visibility, guess what? AI is just going to compound your problems. So I talk about this with strategy, with identifying your constraint. If you don't solve the constraint in your business and you scale, you're just scaling the chaos, you're scaling the problems, you're scaling your constraint, your constraint becomes bigger, harder to solve. That's why when it comes to the fundamentals of your business, the economics of your business, that's what you have to fix. AI is going to be a great tool and a great resource and I highly recommend you implement in your business with the caveats that, uh, I explained here in this episode. Be careful of false positives. Make sure you have the knowledge to prompt, diagnose and interpret. And when you have all this, it can be a super powerful tool in your business but sometimes it's not a replacement for the expertise that exists out there. All right, that's what I wanted to share with you. Go out there, see all the incredible things that AI can do for you and your business and I'll catch you in the next episode. Cheers.