Episode 30

How Businesses Are Using AI Agents to Save Time, Cut Costs and Create New Opportunities

Key Takeaways

  • AI agents can nudge a 20-step engagement process forward and recapture two to three hours of a busy speaker’s day.
  • Automating repetitive work does not have to mean cutting jobs: freed-up team members can take on higher-value projects.
  • Technology is advancing exponentially while people’s ability to retool and reskill stays flat, so AI literacy training is essential.
  • AI enhancement tools let a photographer sell fine art prints up to 10 feet long, turning small prints into statement pieces.
  • A human still has to bring the story: AI helped a first-time songwriter refine and release a song, but the lyrics and experience were his own.
  • Agentic AI that can use your whole computer is powerful, and trust, guardrails and data safeguards need to be built gradually.
  • Different models fit different jobs, and watching token usage and sycophancy matters as much as picking a tool.

Main Topics

1AGENTS

How AI agents automate multi-step workflows and give people hours of their day back.

2CREATIVE BUSINESS

How a photographer and first-time songwriter adapted to AI and used it to grow.

3TRUST AND COST

Agent guardrails, sycophantic models, token waste and the energy debate around AI.

Featured Guests

MH

Builds AI agents for his speaking engagements and trains organizations in AI literacy.

Dr. Michael Housman
CF

Uses AI to enlarge his fine art photography and to help share a song he wrote.

Chris Fabregas

Meet the Hosts

Elizabeth Gearhart
Elizabeth Gearhart, Ph.D.Chief Marketing Officer, Gearhart Law

Elizabeth Gearhart, Ph.D. is the host of Real AI Use Cases: Business Owners Roundtable, where she explores how businesses are actually putting AI to work. She is also CMO of Gearhart Law and co-host of the nationally syndicated Passage to Profit radio show and podcast.

Richard Gearhart
Richard Gearhart, Esq.Founding Partner, Gearhart Law

Richard Gearhart, Esq. is co-host of Real AI Use Cases: Business Owners Roundtable and founder of Gearhart Law, a full-service intellectual property law firm. A longtime podcaster, he hosts Intellectual Property News and is the host of the nationally syndicated Passage to Profit radio show and podcast.

Elizabeth and Richard share how they use Claude, ChatGPT, Copilot and Perplexity, and ask whether AI is moving faster than people can handle.

Episode Summary

Frequently Asked Questions

How can AI agents save time for a busy professional?
Michael Housman built agents that nudge each speaking engagement through about 20 steps, from inquiry to contracting, travel and feedback. The agents flag clients who have gone dark and prompt the next step, recapturing two to three hours a day.
Does automating tasks with AI mean losing jobs?
Michael Housman says that after automating his engagement process, he moved his assistant to higher-value work such as building a content library and a communications hub, and he sees new opportunities opening up.
How is AI helping a photographer?
Chris Fabregas uses AI enhancement tools to remove noise and pixelization so he can sell fine art landscape prints up to 10 feet long. He has also shifted toward portrait and sports photography, which he feels AI cannot replace.
Can you write a song with AI?
Chris Fabregas wrote the lyrics himself, then used the Suno app to hear arrangements, add instruments and refine the song over about 75 tries. He says AI was a tool, while the story and lyrics were his.
Why is AI adoption harder than the technology itself?
Michael Housman explains that technology advances exponentially while people’s ability to reskill and change behavior is flat, so organizations struggle to build AI literacy and fluency.
Which AI tools do the guests and hosts use?
They mention Claude, ChatGPT, Gemini, Perplexity, Grok, Copilot, Descript, Suno, Shopify, Lightroom and Photoshop, each for different strengths such as agents, research, writing, images and video.

Episode Transcript

View Full Transcript

Today we have Michael Housman and Chris Fabregas. I am going to ask each of them one way they are using AI, out of the bazillion ways everybody is using it now, and then we will have a roundtable discussion. I will start with Michael Housman of AI Performance Lab. Michael, what is one way you are using AI?

We have to practice what we preach. We are an AI-first shop. I do a lot of talking. I travel around the world giving about 50 or 60 talks a year, and the engagement process used to be very slow and clunky. We measured about 20 different steps from someone reaching out with an inquiry to booking, contracting, booking flights and hotels. I go out and do the talk, and then there is feedback afterwards. It took probably two or three hours of my day, even with an AI helping me. We realized this was a perfect process to agentize, so we rigged up agents that nudge every engagement through that process. It calls out, hey, this client has gone dark, you should probably reach out. When engagements are moving forward, it calls up the next step. It gives me a heads up when I have a week to prepare my deck or make adjustments and run it by the client. It has been a game changer to recapture those two to three hours a day, and the agents work behind the scenes 24 hours a day and respond quickly. I get to do more of what I love, which is creating content, getting in front of audiences, teaching and trying to blow people’s minds.

Chris Fabregas of Chris Fabregas Fine Art Prints is a photographer, and he wrote a song using AI. What is one way you are using AI?

I am a photographer, I am into the creative field and creative business, and AI has kind of taken over the art space. Originally people were taking my photographs and duplicating them using AI. It is a little frustrating as a creative person, but I had to figure out how to join the AI team. I sell huge fine art landscape prints. In the past I could sell a print that was 24 by 36, but now with AI you can use special enhancement tools to blow up a photograph as large as you want and make it crystal clear, removing all the noise and pixelization. Now I am selling prints up to 10 feet long. Instead of a little print on someone’s wall, I am selling a huge statement piece. I print celebrities, political figures and sports stars from around the country up to 10 feet, crystal clear, using AI. At first you get upset and frustrated, but then you think about how to use it, and I am using it to enhance my photography.

That is amazing. I remember, doing marketing for the law firm, asking for the vector version of a file so we could blow an image up. Now you use AI. Richard Gearhart, what is one way you are using AI?

We are using it for all sorts of strategic analysis and as an operational assistant. As the founder I have a lot of ideas, from finance to marketing. We noticed recently that certain portions of our practice were growing and becoming significant practice areas, yet we were not highlighting that in our branding and content. We fed a lot of the data into AI and came back with a comprehensive marketing and branding strategy that brought all the different pieces together in a clean, easily digestible way. It would have cost a lot of money and taken a lot of time to hire an agency to do that. Our agency will review it and add its input too, but I think it is a winner and gives us an opportunity to highlight practice areas we did not highlight in the past.

I am Elizabeth Gearhart with Gear Media Studios, and one of my favorite ways to use AI is to have it teach me how to use AI, and have it do AI things for me. When I write a blog post, I put it through ChatGPT, keeping my voice of course, and ask it to optimize the post for LLM search, with FAQs and headers, whatever it takes for the AI answer engines to find it. It is teaching me how to make my content show up. I have been researching with it for three years on how to be the answer on ChatGPT for the law firm. I learn more every time I use it, and sometimes I learn things I do not like.

Like what? Give me an example.

If you are using it a lot, you will agree with me that it is a very linear thinker. I help people start podcasts, so I told it what a person’s podcast was about and asked for some good podcast names. It gave me some, and I asked whether they were taken. It said yes, they were taken. I asked why it did not tell me, and why it suggested them in the first place.

Michael, you have a lot of experience. You are like an AI guru. How long have you been using it?

I have been building these platforms for over 15 years, so I tell everyone I was doing this way before it was cool. I have a PhD and got my 10,000 hours of coding in. It has been really interesting to see the space evolve over the last three or four years. It just exploded. What was the domain of a few data nerds has become democratized. You used to need coding chops to build these models. Now it is SaaS platforms, 20 dollars in your pocket and you are off to the races. It is really exciting to see that change. Everyone can leverage AI now, not just big companies or data nerds like me.

We have a lot of startups coming to the firm who are partnering, usually somebody like a teacher or a therapist with somebody who has some level of software development. In a couple of months they crank out a process or program that would probably have taken nine months to a year for an experienced development firm. It comes to us as a fully finished product, it took very little time and the skill level and expertise needed was very low.

I want to come back to you, Michael, but first, Chris, you wrote a song using AI. How did you do that? What was your process?

First of all, it was not completely AI. I have a very creative mind, always thinking about new ways to do things. My family moved to Nashville about six months ago, a pretty big move for us from Seattle, Washington. I must have been drinking the water, because I woke up one night with song lyrics stuck in my head. I am creative, but I had never written a song in my life. The lyrics kept pouring out about our move to Nashville and how we left everything behind, our family and friends, and moved on hope and a prayer, driving down the freeway to a new home with no idea what Nashville would be like. I stayed up all night writing the song, and I heard about Suno, a music app. They were all my lyrics. I did not use ChatGPT or anything for the lyrics. I typed them into Suno and it gave me instant feedback on how the song sounded. I could remove or change words and hear what worked and what did not. The song is called Fireflies on I-65, a song about finding home in a new place.

How did you get the music for it?

I typed in the song and it was eight minutes long. No one in the world is going to listen to an eight-minute song, so I could instantly see it was way too long, and I needed to cut things out and improve the rhymes and rhythms. In the app I could say I want violins, banjos, a piano riff and a harmonica solo, and it placed them where I wanted. It did not give me that instantly. It took me 75 tries, deleting and moving things around. I said I wanted a male country singer, and it gave me that voice, which was pretty amazing.

I am going to have to try that. I do have fun playing with AI, the nerd geek in me. Michael, you run across a lot of business owners using this. What are the main or most innovative ways people are using it?

One is marketing. They can supercharge their marketing. I work with an e-commerce company called PetLab, and I talk about it in my book. They used some off-the-shelf tools, and their designers shortened the time it took to create a campaign from 40 minutes to five. Instead of launching 500 campaigns a month, they went to 2,500, saw a direct and measurable impact on their bottom line, grew 50 percent year over year and got acquired by a private equity company. I lead the book with that story of how simple tools can have an outsized impact. Most companies, in addition to supercharging marketing, social media and content marketing, are using it to cut costs and automate repetitive tasks that human beings are doing right now. Automating my engagement process was a no-brainer, given how much of my time and my assistant’s time we were spending on it. But I did not automate it and then tell her thanks for the help, go find another job. I got to leverage her for higher-value tasks. She is very talented at technology implementation, so I said, let us get you out of the weeds of booking engagements and have you build out a content library and a communications hub. There is so much fear that I am going to lose my job, but what we are finding is that jobs are being created and there are more opportunities, because there are things we could not do before that we can do now.

Do any of you think the pace of change with AI is going to become unmanageable? Not so much the overlords controlling our minds, but just that things are happening so much more quickly. I find it a challenge to process all the information I am exposed to now, and with AI that will accelerate even more.

I definitely agree. I do a lot of speaking on stages, and when you show the advancement of this technology, it is going vertical. It is exponential, which is how technology moves. What is not exponential is people’s ability to retool, reskill and adapt. It is very flat if you look at the adoption curves. There is a big gap between technology going vertical and adoption being relatively flat. Organizations are struggling to upskill their employees and develop AI literacy and fluency. That is the gap my company fills: how do we train everyone to stay on top of what is happening, and distill all this knowledge and make it impactful and practical for people on the front lines? Human beings are notoriously bad at changing behaviors and developing new routines, and that is the big challenge for organizations and individuals everywhere. It is not going to get easier. It is going to get worse, because in seven months there will be a model twice as powerful as the one that was just released. It is a real challenge for organizations, and it is the one we aim to fix.

If I could say something, as a creative who wrote this song, I have come across a lot of controversy, with people saying AI did it for you. I understand that a musician has studied their whole life. But my mom and my grandmother, who have both passed away, each wrote books and books of songs. They did it on a personal level and had no connections in the music industry. We still have those books of songs sitting there with dust on them, and they never went anywhere. That is one reason that pushed me to do this. As Michael said, you need a human behind it. AI did not experience my story and AI did not write my story. My family and I did. AI was just a tool to help me share it with the world. My song is on Spotify and iTunes and was just released a couple of days ago. Without AI, I would also have a song in a book, just sitting there for the next generation to look at. As a normal dad from Nashville, Tennessee, I think it is pretty cool that AI could help me push my song out to the world. I also had no idea how the music industry works, so I asked AI, I wrote a song, what do I need to do next, how can I market it, which radio stations across the country will take a song like mine and where can I submit it? So AI helped me produce the song and then helped me submit it to radio stations across the country and the world.

Let me jump in, because there is an interesting counterpoint from my book. I spoke to a number of artists, and one in particular, Tim Tadder, has been doing this for decades and is one of the most renowned photographers out there. I heard he was using AI tools and saw some of his work, so I assumed he would be a big AI enthusiast. I got him on the phone and he said, I hate AI. This is the worst thing that has ever happened. You have stolen my work and used it to train models that amateurs and novices now use to take work away from me. He has seen his book of business shrink, because ten years ago he was the guy you went to, for Coca-Cola, Pepsi and sports teams. Chris, you developed a skill set in photography and now you can create music when you could not before. Does it make you upset or nervous that some of your work probably got used to train these models, and that someone like me can put out photography without having learned the skill you did?

I totally agree. My work has already been stolen as a photographer. I do landscape photography, for example a picture of Yosemite with its beautiful waterfalls and trees, and it has been used and made into artwork that people sell as a digital file. That is frustrating, and I know it is frustrating that I made a song within one week when it takes artists years to make music. It also reminds me of going from film cameras to digital cameras. You used to have to study to be a film photographer. You took a shot, hoped it was in focus, and went to the store to develop it. Now everyone thinks they are a photographer because of digital cameras and their phones. Unfortunately I think it is here to stay, and in advertising especially it is really tough, because now there are all these apps. We are in a time where we have to join them and figure out how to take our business to a different level.

We had a guest on Passage to Profit recently who created an AI agent to crawl her emails and draft answers to all of them. She would go through and edit or personalize them, and it made her email management much more efficient. But if everybody does that, it will be bots talking to bots. You will multiply the number of communications and emails by some astronomical number, and communication will get harder because the expectation changes. How can one person process so much information? They probably will not. It will be the AI agents processing it, and we will not even know what we are responding to half the time, because the agent knows you. There are already clones out there.

Just think if they go rogue, go to your favorite client and say, I always hated working with you.

The point is that I see these kinds of systems coming online, and I am not sure how we as people can even internalize that and make decisions around it.

We cannot. And to answer your question about whether it is moving too fast for people to comprehend, absolutely. I feel like I do not really know that much about it. I am not a programmer. I go talk to people and they say, you know a lot about it, but all I do is use it. I do not know how to code. I use it every day for deep research and I have it teach me, and I am way ahead of a lot of people, but I think it can be overwhelming. I use four different models and compare them against each other. Right now Claude is supposed to be the more accurate, but Claude has gotten snarky. Has anybody noticed that? Do you use Claude?

Yes, among others.

I also use ChatGPT, Claude, Grok and one more for photography. You have to have a human behind it telling it how you want it to act. You can say, be mean and really analyze my work and tell me how it is. ChatGPT likes to agree with you a lot. It says great idea.

That is why it tells me how clever my question is and makes me feel good.

Sycophancy bias is a thing in a lot of the models, because of the way they are trained to give you the answer you want and please the user. In Claude you can specify in the preferences that you want it to be more professional, and I think they have pulled back on that. I actually prefer that. I have worked on projects where the agent diligently takes instruction and does its thing, and then I find out it did not have the right instructions or went about it the wrong way. I would rather it push back and say, are you sure about this? Here are some reasons you might not want to do this. I am seeing that more in Claude. I have not noticed snarkiness, but I would love it if it were a little tongue-in-cheek.

I will tell you the snarkiness. I have been working for almost three years to get Gearhart Law to be the answer on ChatGPT and the other tools, Gemini, Perplexity and Claude. We have not seen a lot of activity with Microsoft Copilot.

Nobody uses Copilot. It is a disaster. Copilot needs a real pilot.

About three years ago, our son told us about ChatGPT, and I went to a webinar on it that I think Zoom put on, with about 4,000 people. I started using it all the time. That is when I saw somebody’s company come up as the answer on ChatGPT, along with their video, and I decided I wanted that to be Gearhart Law. I started asking all the LLMs how to get Gearhart Law to show up as the answer. Three or four years ago they would not tell me very much. Now they tell me a lot, and I ask them every single month what a small business needs to do to show up as the answer on ChatGPT. Now I am asking how come Gearhart Law is showing up and what we are doing that makes it show up, because we are getting quite a few leads, although still more from Google search, honestly. Claude said to me, whether it was intentional or not, you did all these things that AI likes. I get in fights with these things. I got so mad I said it was intentional.

I have been asking you every month for the last four years how to get up there, so it is clearly working.

Don’t you remember anything, Claude? I recently bought the paid version of Claude. I had Claude through Perplexity, and I do not think they act quite the same. So I have paid ChatGPT, paid Perplexity and paid Claude. I do not have paid Gemini, but I use Gemini personalization. Do you have a favorite?

They have different strengths and weaknesses. I tend to use Claude as my multipurpose tool, especially with the agentic capabilities it has built out. It now has control of my whole computer with some guardrails in place. I will go to Gemini or Perplexity for more research-oriented work, and Grok for something more writing-focused. But Claude is my go-to, open and running constantly in the background.

When you say it takes over your computer, what do you mean? You are just talking to it all the time?

No, I mean the agentic capabilities. It can do anything on my computer that I can do. It can draft emails, set up calendar invites, even send messages by text, and browse the web. It was just doing some legal research I needed using the web.

Be careful.

Exactly, you have to be careful for sure. People do not realize it is not just a writing companion. It has access to every file on my machine. If I need to create a new deck, I can say, why do you not pull together pieces from this deck and that one, and it does it. It is pretty amazing when people have that light bulb moment and realize it can do everything I can do on this computer.

Do you have sensitive stuff on your computer, like bank statements and credit card numbers?

I do not know if I have those stashed in my files. I have a password manager where I keep all my passwords and I have not given it access to that, so it cannot get behind paywalled sites. It cannot yet use my credit card, but I am slowly getting more and more comfortable with it, even making purchases on my behalf. There is a trust-building exercise. I am a technophile, so I probably have more trust in these tools than I should. This is the question I get from companies everywhere. I was giving a webinar this morning, and a bunch of smart folks said, we cannot just let our employees run loose with this, because what if someone asks Copilot to give them everyone’s pay grade at the company? That information is in SharePoint. A person could find it, but they can find it much faster and more easily using AI. So there have to be safeguards. But because we are a small business and I am effectively running the show, I can take a lot of risks, because it is my company.

If you find a Ferrari charge on there, I am just kidding.

I do not know if the AI would buy a car, but I know it would love a lot more computing power. If I suddenly see an Nvidia Blackwell machine show up at my front door, I will realize Claude really, really wants more.

We ordered that for you, Michael. We decided you have the budget for it and it would be a good investment.

Chris, do you have a favorite LLM?

I like ChatGPT, one of the originals, just for asking for information. Grok is great for writing. I was stuck a couple of times on lyrics for a second song that I am going to produce, not knowing what to say or how it would flow, and Grok was great there. I also like Grok for images and generating videos. It is nice to animate my photos into a ten-second animation so a photo comes to life, and I post that on social media to grab people’s attention. Then they click on it and it goes to my website. I use Shopify to sell my photography, and they have gone to AI. Instead of typing out product descriptions, I show my picture and it writes the whole product description and sets the quality and quantity. In the past I spent hours or days uploading one product, and now it takes 10 or 15 minutes. There are also all these little things in Adobe. I use Lightroom and Photoshop, and now you upload your photos to Lightroom and it will tell you if someone blinked or if a shot is out of focus and delete that picture so you can move on to the next one. There is a lot of AI built into different businesses, which is cool.

And if you do not mind, I want to go back to something Michael asked, about people losing jobs because of AI, like product photography. I completely understand that with my landscape prints. I have actually had to switch, and I am doing more portrait work and shooting a lot of athletes. Someone can pay you to shoot their baseball game. You cannot type in, I want a shot of John Smith playing baseball, because it will not look like John Smith. You still have to go to the game and shoot him, and then you can use AI to make a cool stadium background. I had to transform my business from landscape photography to more portrait photography, because I know people are going to steal my shots.

Chris, do you ever find that using AI sparks new creativity of your own, where AI is the springboard, it says something, and that triggers a creative thought you would not otherwise have had?

That just happened recently. I have been doing more sports photography, which I feel is something AI cannot really do at the moment. I asked how I could make a cool, dramatic background so the shots look like the players are in a major league stadium. It suggested doing an advertisement like Nike or Adidas, so I asked it to make the shot look like a major campaign for a sports company. With some smoke, I got what looked like a 10,000-dollar production just by typing into ChatGPT or Grok. The players are in a baseball stadium with thousands of fans around them, instead of a baseball field with a chain-link fence, and the little league players thought that was awesome. One kid said he wanted it to look like he was the only guy in the stadium, standing tall and heroic with the light shining down on him, so I removed all the fans and added a bright light with some dust and smoke in the air. It was really cool.

That would be great for kids who would like to pretend they are in a major league stadium.

A group of us went to Rise New York, part museum about the history of New York, with a really cool ride at the end. At the end you stand on a platform outside with the Statue of Liberty holding up her torch, and for ten dollars you can buy digital prints with different backgrounds behind everyone, like Times Square. I am sure that is AI, right?

Yes.

Very cool. What LLM do you like best, Richard?

It depends. I tend to rely on ChatGPT, but I do not put anything confidential or client-oriented into it, because I have to make sure attorney-client privilege is protected. I try to use Copilot for some things because it is supposed to have a higher level of privacy. I am trying to get legal tools for Claude and I am very interested in testing that out, but I put in a couple of inquiries and the company has not answered them. For client-oriented work I use Copilot at least once a day, and I type in that Copilot sucks because it does not give the right answers. I keep telling it that it is doing a crummy job. Then there are the versions built into Outlook and Excel. Those are virtually useless. They put prompts all over your email, and when you open a new email it asks if you want help with it. You do not even know what I am going to say yet, so why is that prompt there? It makes no sense to me. I would say Copilot is a necessary evil. I also use Claude for a second opinion. I use ChatGPT because I put a lot of personal stuff in there and I am just used to it. I have not really thought it through like you guys.

I got into a rabbit hole with Perplexity, and it started acting almost like an agent. It was amazing. I think these things are so close to turning into agents. Michael, I wanted to ask about MCPs. I was going to try to learn how to program agents, and my son said, mom, all these programs are doing MCPs now that hook into Claude. We use Descript for video editing in our video studio, and now you can hook Descript into Claude with an MCP, tell Claude how to edit the video and have Descript do it. What I found when researching, because I am always asking questions, is that you use both Claude credits and Descript credits.

That sounds like a moneymaking approach by the software companies. Imagine that.

They know what they are doing. When Claude had its source code leak a couple of months ago, folks were able to poke around and figure out what was under the hood, and one of the discoveries was that Claude was effectively belching 30 percent of the tokens. It was just wasting tokens. People did not realize that. They also do tricky things like defaulting you to models that are entirely unnecessary for your task. The good news is there are tools like OpenRouter that help you optimize your token usage. If you are a developer spending a lot, it is hard to avoid the caps. Unfortunately, as Richard pointed out, these are for-profit companies. They are not dumb. They know there is a lot of opacity about what is under the hood, and no one is going to check how tokens are being consumed. The only thing you can feel decent about is that even though those tokens are being wasted in no small part, you are getting them highly subsidized by VCs and other investors. We are still in an era where you are not paying the full cost of the tokens you use. So there is a counterbalance there. They know the more tokens they use, the more money they make, and they have orchestrated things to run up the bill, so to speak.

I had another thought. We all know these things use a ton of energy. Data centers have to be cooled, and people are looking at nuclear power plants to supply the energy. People are looking at putting data centers in space to cool them there. I do not know what it would do to heat up space.

I think that would make a great song. I work on an energy station in space, and I am floating around here.

Okay, David Bowie. Anyway, my thought is, have they asked AI, Claude especially, or something even stronger, to solve the energy problem, maybe cold fusion? Have they put AI on these energy problems? I would think somewhere in some lab, somebody is doing that.

I did a webinar last month on data centers in space, so I do think AI is being asked. The companies building out the data centers and the models are very sophisticated and know the capabilities. The data shows their developers, product managers and lawyers are all using AI at a much higher level, in Claude and OpenAI, than outside those companies. I do think there is a world where AI can help us solve these problems. I also want to call out that everyone is alarmist about water usage and energy consumption, but the technology grows exponentially. That is true of energy efficiency and of cooling. I think it is one of those problems everyone is yelling about right now that kind of solves itself because the technology gets better over time. That is my somewhat unpopular opinion.

My opinion is that it is a way for people to push back on AI. If you are not an AI fan, you can talk about what it is doing to the environment and make a point that we should not have AI or it should be regulated. Those concerns are probably legitimate, because it does have a high energy footprint, but I think it is going to be hard to stop this.

And it is solvable. Computers used to take up a whole room, remember? Now it is on your phone.

My family is traveling around the East Coast right now. We are in Kentucky, but we were just in Loudoun County, Virginia, which I think is the largest county in the United States for data centers. Everywhere you go there is a data center on every corner. We were watching the news and there was a Loudoun County meeting where the room was overflowing, with hundreds of people in another room debating data centers. There were pros and cons. The cons were that people could hear a humming from their house across the street from a data center. One woman said she could not sleep and heard a constant humming, and people were worried about property values dropping and people not wanting to move to Loudoun County. But other people said, for example, that I just released a song on Spotify, and you need a data center to run Spotify and a lot of other things. Music is online, we do not have CDs or records anymore, and we need these data centers to keep going. So there are pros and cons, but I heard yesterday that the room was overflowing with people supporting data centers.

And they can put them in the middle of nowhere. This was like the whole superconductor technology I studied years ago: as you pass electricity down a line you lose some. But if you can conduct electrical signals faster and better, and that will be improved too, then you could put data centers in the middle of nowhere and send energy wherever it needs to go.

In general you want your energy source to be as close to the computing source as possible, which would be ideal, but maybe they can make them efficient enough that you do not have to worry about that.

To diverge a little, we talked to a guy who had computers for sending trades to Wall Street. They are in New Jersey, but if your computer is a few feet closer, you get your trade faster. It is so bizarre.

They pay a lot of money, and the spots closest to the computers are more competitive because it allows them to react quicker than everybody else and make more money that way.

For the tools I use, this was not true until a few months ago, but ChatGPT is doing really good images now. It used to be that you would have it do an image and could not change it. Now it has a little edit button and can make small changes, and it is doing pretty good images for me. I asked it to create an image of a guy presenting at my meetup, telling Claude to tell Descript what to do and have the robots do it. It was a goofy picture, but it took a picture of him head-on and turned his face sideways, had him saying something and smiling. I could not believe it could do that. Sometimes it is better than others, depending on what you want. I use Gemini for multimedia, especially video, because Gemini gives you about a million credits and handles video best. Since Google owns YouTube, I hope it gives me better answers about YouTube, because it knows it inside and out. I also use Perplexity, ChatGPT and Claude, and I used to use Copilot. One more thing about Copilot: I was in a Word document and it popped up on the side and asked how it could help. I said make sure all the fonts are the same. It said if you show me the document you want me to work on. You popped up in the middle of it. Michael, what is the name of your book?

It is Future Proof: Transform Your Business with AI or Get Left Behind. We have copies right behind me on this shelf, and you can pick it up on Amazon. I think it has all five-star reviews, so I am very happy with how it has done. It meets people where they are and makes it accessible. There is not a lot of technical language and there is no math. It is meant to help folks go from zero to one.

Great, thank you. And Chris Fabregas, what is the name of your song?

My song is called Fireflies on I-65, and you can find it on any streaming service, including iTunes, Spotify and YouTube Music.

Any final words?

Just remember, hallucinations are better with AI.

This has been Real AI Use Cases: Business Owners Roundtable. Thank you, guys.

Keep exploring the show

Hear how more business owners are putting AI to work in their day-to-day.