How Businesses Use Multiple AI Agents to Build Software

[00:00:00] Richard Gearhart: But if we're even still here in 2027, isn't AI going to kill us all? 10% chance that we're all going to be exterminated by AI [00:00:08] Elizabeth Gearhart: it gave me a couple things it could do to kill us that I don't really want to talk about here. [00:00:11] Richard Gearhart: The fact that it had an answer is scary. [00:00:13] Todd Drowlette: In a matter of three and a half hours it did what used to take us three weeks of three people full time doing. [00:00:18] Sam Desany: You can easily spend $3,000 a week on Claude code with it just running 24 7. And then you ask yourself, how much am I actually saving? [00:00:27] Brandon Bellamy: I think we're making our own worst enemy. But again I think that it's a tool that should be used. [00:00:32] Elizabeth Gearhart: This is real AI Use Cases Business Owners Roundtable. Here's how real companies are using AI right now. So now it is time for real AI Use Cases Business Owners Round table. I'm going to pick on three of the people here and ask them how they're using AI in their business. And then they'll give me one way and then we'll do a round table where we discuss AI. And there's a lot to talk about these days with AI So I'm going to start with Todd Drelet. What is one way you're using AI? [00:01:03] Todd Drowlette: I hired Sam and Sam is running the AI for us. [00:01:07] Elizabeth Gearhart: That's what a lot of companies do. [00:01:09] Todd Drowlette: An example I can give because he does a ton of things with AI So we used AI to streamline. I have one tenant called Label Shopper. Peter Harris Clothes are an off price clothing chain much like a TJ Maxx or a Marshalls. And I work with them in 17 states. And when we say hey, we're going to go out, we need six stores for back to school. We used to have to physically go search the Internet, pull up markets we want to be in. It would take us 3 people full time for 3 weeks to identify potential locations using AI Sam wrote basically a program that would go out, search the Internet, pull a bunch of marketing brochures, run demographic reports all in real time in a matter of three and a half hours. It did what used to take us three weeks of three people full time doing. And then things like Letter of Intents Lois. It will auto complete contracts. It's really good at doing that. So that saves a ton of time. So we've been doing this and those [00:02:04] Richard Gearhart: things are so tedious to do too. [00:02:06] Todd Drowlette: That's one of the things that I trust AI to do where it's the tedious stuff that you can just scan it and go, oh, quick fix. [00:02:12] Elizabeth Gearhart: So let's go to Brandon Bellamy. So what is one way that you've used AI recently or your favorite way? [00:02:19] Brandon Bellamy: I like to use AI AI to identify opportunities. I use it more of a lens than a crutch. Right. Right now I'm selling a building in Syracuse. Right. For easy math, let's say that building is 19,000 square feet. I would use AI, make a project, a sub chat, something like that. And I'd upload the local assessment report so I have all the tax maps for the whole city of Syracuse and I'll compare that building with or I'll run that report and find buildings that mimic the building that I have and contact those owners to see if they're interested in another 19,000 square foot building. [00:02:53] Elizabeth Gearhart: Wow, that's really smart. We get a lot of repeats on this show. I've never heard that one before. [00:02:59] Brandon Bellamy: I just wanted to backtrack and you were talking about the sources and you know, me not using it as a crutch. The last thing I want to do is ty ask me where I got this information from or a landlord or a client and I don't have a real life source for that. I can't just say Claude popped this out or chatgpt popped this out. [00:03:16] Richard Gearhart: So. [00:03:16] Brandon Bellamy: So again, I'm big on the using it as a lens and not a crutch. [00:03:19] Elizabeth Gearhart: Okay. Sam Desney, what's. You're the AI guy, but what's your favorite way to use AI? [00:03:25] Sam Desany: My favorite way is probably just to code some of the sites that we're working on, the front and back end of the sites. Because my background in school is a lot of marketing and data analytics. So a lot of the stuff that I have Claude and I have Codex or lovable doing is they code the back end and the front end based off of complex prompts that I give it and it gives me feedback and I can send it straight to the Internet or I can Text Todd at 2:00 clock in the morning, which sometimes I do with a finished product saying, how does this look? [00:04:04] Elizabeth Gearhart: But I want to do bring one thing up. Yeah, it's doing the coding, but it's really your prompt that is making the difference. [00:04:11] Sam Desany: So the system that I use is I actually have four different AI agents. So I have my brainstorming AI agent that I send my half formed ideas to and then that agent talks to one that's already automatically connected. They create prompts that are sent to the third agent and then that Agent gives me prompts and feedback and gives me the specifics that I need to send to my actual coding backend and front end agent. [00:04:42] Richard Gearhart: Can you go over that again and bring it up a little bit? Because what you said is very smart and very technical, but our very high level is very. Try to explain it in more general terms. [00:04:54] Sam Desany: So let's just say, for example, there is a person who is coding, who's your employee and you are a manager at this firm, but this coder only really speaks in coding language that you can't really communicate to them. So say this manager has their ideas, they're coming into the office and they talk to their team, which are two, maybe three other employees, and they're like a translator. They take the manager's half brought to life ideas like all managers, but they take those ideas and they digest them and then send them to the coder in a way that the coder can understand. So the coder doesn't have to do the back end and the front end and understand these half put together thoughts. [00:05:45] Richard Gearhart: Yeah, that's one of the important issues is that we work with an AI company. And I'm not always sure that the person who's actually working on the project really understands the context of how it's going to be used or why it's being used. And so if you have a way of communicating that bigger picture, I think that helps the coder and then it ultimately makes for a better project. [00:06:09] Elizabeth Gearhart: So, okay, we're going to come back to you because we're going to have a little bit more discussion. But Richard Gearhart, what is one way you're using AI? [00:06:16] Richard Gearhart: One way. Just this last week I put our financials into Copilot and I got a valuation for our business. [00:06:25] Todd Drowlette: Is this your official announcement of your retirement? [00:06:28] Richard Gearhart: No, not yet. [00:06:30] Elizabeth Gearhart: I'm not letting him retire. He cannot sit around the house. No, I'm not retiring. I'll never retire. [00:06:36] Richard Gearhart: But I thought it came up with a fairly realistic number one that I had and gave a range. Right. And then also it gave me suggestions on how to improve the valuation of the business, which I took to my business advisor and immediately said it's all wrong. But actually I thought some of it was pretty good. [00:06:55] Elizabeth Gearhart: I would choose a second LLM and ask, but I know you don't want to put anything through a non enterprise version. [00:07:01] Richard Gearhart: In the past you would have paid somebody what, $15,000 to do an evaluation and would have taken two months and they would have interviewed everybody in the firm and done all sorts of Stuff. And if you're just looking for a back of the envelope type of number, it works pretty well. [00:07:16] Elizabeth Gearhart: Yeah, for me. So I decided I want to take all the content from this podcast and have AI help me write a book. And Richard will be a co author, of course, because he's here gathering the content too. So I asked AI, because you got to understand, to post the podcast. I get the transcript and I put it through ChatGPT, and I'm going to automate this process. I'm working on doing an agent for this and I say, okay, chatgpt, generate the description for Kastos, which is where we upload our podcasts and distributes it everywhere. And make it SEO and aeo, optimize it for search engine optimization and answer engine optimization, give me FAQs, give me headers, et cetera. I'm sure that Sam knows everything I'm talking about. I've been doing it in chat. Chat has all the transcripts already. So the first step is to contact everybody, make sure they're okay with being in this book. But secondly, I think I might actually put them in Gemini and say, okay, give me an idea, give me a book, and just see what it comes up with. It's still gonna take a lot of human intervention and a lot of work. [00:08:25] Richard Gearhart: You're just saying that. Cause you want the copyright. [00:08:27] Elizabeth Gearhart: Yeah. So to get to the point, okay, I asked ChatGPT what should I call this book? And it said, you should call it Real AI use Cases. So it matches the podcast. And then it gave me. But you need to have a tagline. You gave me this big old long tagline. And so Richard and I were talking about it, and Richard actually came up with what I thought was a very good tagline for it. So this is what it's going to be. It's going to be real AI use cases. 50 ways you can use artificial intelligence to improve your business. [00:08:57] Todd Drowlette: People love numbers. [00:08:58] Richard Gearhart: Yeah, they do. [00:08:59] Elizabeth Gearhart: Yeah. And that was Richard's idea. [00:09:01] Brandon Bellamy: Spoiler alert. [00:09:02] Richard Gearhart: We might include 51 ways extra value. [00:09:06] Elizabeth Gearhart: We're going to put 2026 edition because we want to try to do this every year. I kind of feel like it's a documentary of how businesses have evolved using AI. [00:09:15] Richard Gearhart: But if we're even still here in 2027, isn't AI going to kill us all? 10% chance that we're all going to be exterminated by AI. [00:09:24] Elizabeth Gearhart: I asked Gemini, I said, how could it kill us? It doesn't have hands, it doesn't have arms. Like, what's it going to do? What's it going to shoot us with? And it gave me a couple things it could do to kill us that I don't really want to talk about here. [00:09:34] Richard Gearhart: The fact that it had an answer is scary. [00:09:37] Elizabeth Gearhart: Yeah. You have to put gaps in processes. You have to treat AI like you treat nuclear weapons, like the nuclear code [00:09:44] Brandon Bellamy: called a human at the helm. That's the term you're looking for. [00:09:46] Elizabeth Gearhart: What do you think? [00:09:47] Brandon Bellamy: I think that we're creating all these things. I think we're making our own worst enemy. But again, I think that it's a tool that should be used. I don't think we should rely on it as heavily as we are. Like, even with your financial information, you want to be careful about what you put in there because you get your chatgpt wrong one day and they expose everything you got going on. [00:10:05] Richard Gearhart: Yeah, that's the risk, right? [00:10:06] Sam Desany: Yeah. You look at these big companies and it's actually quite recently, the past couple months, and they're starting to hire people back that they let go because they had all of these AI and these software systems. And I know that everybody saw a year, maybe two years ago where they were letting all these software engineers go, but now they're actually hiring people back because they're realizing that not only is it hard to manage, but it actually costs a lot more money using AI, using tokens and credit usages for big things like CLAUDE code. You can easily spend $3,000 a week on Claude code with it just running 24. 7. And then you ask yourself, how much am I actually saving? Whereas I could just have kept the person who was working for me before. [00:10:54] Elizabeth Gearhart: I saw that companies weren't realizing the ROI that they thought they would with AI. But that's a really good point. So we talked to somebody who's like, oh, you can have Claude do your video editing for you. It can tell descript what to do. But you have to buy credits from CLAUDE and from descript. [00:11:10] Sam Desany: And you still have to prompt it. [00:11:12] Elizabeth Gearhart: And you still have to prompt it. [00:11:13] Sam Desany: It's not just thinking for itself, saying, hey, here's a suggestion. You have to come up with an idea. It's not a person. And I think it's a great idea to use it as a tool. It's better used like that. It's. And I submit to you that if we have people who are operating it and don't do stupid things like give it nuclear codes or give it people's financial information, then you'll be fine. Use it for tasks that are Repetitive that you don't need a human's brain to do. [00:11:43] Richard Gearhart: I think the danger from some bad actor using AI to commit atrocities is much higher than AI just trying to take over. One of the early AI cases I remember was in the biological sciences, where a research lab in Europe created over 50,000 compounds using AI. These compounds were toxic to human beings. A lot of them weren't even detectable using current medical tests. So if somebody really wants to rain havoc on the human race, they can just make a poison and it's the poison. I still think the probability of that happening, not that I'm trying to give anybody ideas, is much higher than AI just deciding, one day I'm going to exterminate the human race. [00:12:31] Brandon Bellamy: I agree. I find even with my chats that sometimes the AI kind of gives its own opinion or inserts or is giving me back or telling me what it thinks I want and what I want to hear. So when it comes down to kind of like making decisions that. All right, let's say you want to spend $100,000 of that company evaluation on a Bentley. I like Bentley's. You deserve a Bentley. [00:12:51] Richard Gearhart: Right, but can I get one for 100,000? [00:12:54] Brandon Bellamy: What if your AI project manager or whatever kind of chat you're using doesn't think that's a good idea and blocks your card? Then you're like in a very negative way. [00:13:01] Elizabeth Gearhart: And even for research, it's iffy because I put together some questions you could ask ChatGPT to see if it recognizes your business, how your AEO marketing is doing basically. And I tailored it for Gearhart Law and I kept asking and Gearhart Law kept coming up as an answer, which was great, but specific to you. [00:13:21] Todd Drowlette: Now that people are searching it, even [00:13:23] Elizabeth Gearhart: though I was saying to it assume I'm a small business owner looking, I still think it's trying to give me positive strokes. So I'll keep using it. [00:13:32] Todd Drowlette: The way you phrase the question has a ton to do with and then it'll give you what you want to hear. [00:13:37] Elizabeth Gearhart: So I think you have to have somebody else ask it that you've never interacted with before. Right. [00:13:42] Richard Gearhart: Sometimes I find myself just going to the AI without even spending time thinking about the issue myself. Right. And so I have to watch myself because I don't want to lose my own ability to solve problems and think about things first and then go through and verify it with AI. So it's dangerous. You can just start going to AI for every little thing and who knows what the consequences of that are long [00:14:08] Elizabeth Gearhart: term but it's still wrong a lot of times. So I use different LLMs and compare their answers to each other and sometimes they're all aligned, sometimes they're not, and they give different answers. But I have my website on Manus and they had to take it down, put it back up, and it worked. So my website's back up. But I asked Claude, I said, can you tell me if Manus is operating normally? And Claude said, it's been doing great for 90 days. There have been no disruptions in Manus for 90 days. And so I said to it, I said, really? You mean like having to pull all its data off of Meta and then re establish its data on its own servers in its own company is not a disruption to you? And it was like, oh, sorry, I didn't look at that part of it. [00:14:54] Sam Desany: Yeah, it can be bad to go down a rabbit hole like that. Which is another big reason why I have those four agents before anything actually gets put into code. Because 99% of use cases, they correct their own mistakes. And even if they don't, I verify the prompts before I send them in. It's more of just a faster way of doing what I'm already doing. It's not creating something that I don't necessarily know how to do. [00:15:21] Elizabeth Gearhart: So for people who don't know what agents are, they'll just do a series of tasks for you, right? [00:15:27] Sam Desany: Agents can work in one chat at one time. So say you give a prompt, you can work from one agent, which is doing one task from that prompt. Or simultaneously, you can work 100, a thousand, 10,000 agents at one time. They're like employees. [00:15:45] Todd Drowlette: But the thing to remember is the most important agent you'll ever have is your commercial real estate agent. [00:15:50] Richard Gearhart: Oh, I was going to say your secret agent, your secret agent, or your [00:15:56] Brandon Bellamy: patent agent [00:15:59] Elizabeth Gearhart: with that. Thank you very much, everybody. This has been really Real AI Use Business Owners Roundtable. [00:16:05] Brandon Bellamy: You have been listening to real AI use cases Business Owners Roundtable with hosts [00:16:10] Elizabeth Gearhart: Elizabeth Gearhart, marketing expert and PhD researcher using AI every day. [00:16:15] Brandon Bellamy: And Richard Gearhart, entrepreneur, business owner and [00:16:18] Elizabeth Gearhart: intellectual property attorney specializing in innovation. We hope you found this valuable. Join us again for more stories. [00:16:24] Brandon Bellamy: Because the future of business is driven by AI.