AI and Me: What Could Possibly Go Wrong?

What happens when you combine decades of business experience with artificial intelligence? Sometimes the results are surprising—and sometimes they're downright frustrating!

In this episode, Roger Pearson shares his firsthand experience working with ChatGPT to rebuild the Seagull Technologies website. What started as a project to consolidate two websites evolved into creating custom tax planning calculators, an online business education platform, and integrated payment and scheduling systems.

Along the way, Roger discovered both the remarkable possibilities and the limitations of AI, including the importance of testing results, solving problems one step at a time, and knowing when to stop chasing a solution.

This isn't a discussion about what AI might someday accomplish. It's the story of what one small business owner actually accomplished by combining his knowledge and experience with AI technology.

**Part One of a two-part series** exploring how artificial intelligence helped transform a business website—and the unexpected challenges that followed.

Questions Answered
How can small business owners use ChatGPT to improve their websites?
Can artificial intelligence help create custom software without professional programming experience?
What are the advantages of developing custom website features instead of relying on existing software?
How can AI help businesses develop tools that solve real customer problems?
How can AI help automate business processes such as payments, customer intake, and appointment scheduling?
What are the biggest challenges and limitations of using AI for website development?
Why is it important to test and verify software created with artificial intelligence?
How can business owners troubleshoot AI-generated solutions more effectively?
What role does human experience and judgment play when using AI to develop business solutions?

Choose How You Enjoy This Episode

Watch the video, listen to the audio, or read the transcript. The choice is yours!

Watch the Video

Listen to the Podcast

Prefer to Read?

Read Full Transcript

AI and Me: What Could Possibly Go Wrong?

Roger: I have to tell you about something I've been doing over the past several months because it's turned into quite an adventure. And to tell you the truth, when I started, I had absolutely no idea where it was going to lead, but it's quite a story, so I hope you stick around. 

Announcer: You are listening to the Mastering Business Basics podcast, where we discuss how to build a solid foundation under your small business to improve your chances of success. And now, here is your host, Roger Pearson. 

Roger: Now, I've been building and maintaining websites for quite a few years, and if you've ever done any of that yourself, you know it can be an exercise in frustration. 

You get an idea for something you'd like your website to do. Sounds simple enough. So you start looking around for a piece of software that'll do it. In my case, I use Drupal to build my websites, and it has thousands of these little add-on programs called modules. You can find modules that do just about anything. 

Well, almost anything. The problem is you find one that looks like it'll do what you want, and then you discover it doesn't do quite what you had in mind, so you look for another one, and maybe that one requires two other modules to work, and then you discover that one of those hasn't been updated in three years, or it doesn't work with the version of Drupal that you're running, or you finally get everything installed, spend a couple of hours configuring it, and realize it doesn't do exactly what you wanted. 

Been there, done that more times than I care to remember. And that's pretty much how I'd been doing things for years. But I decided to try something different this time. I started working with ChatGPT. Now, I'd used artificial intelligence before for research, but I hadn't really thought about using it to help me develop custom software for my websites, and that's where things started getting really interesting. 

Instead of spending hours looking for somebody else's software and trying to make it do what I wanted, I could tell ChatGPT what I was trying to accomplish, and it could actually write the code to make it happen. Now, before anybody gets too excited, I need to tell you something: It doesn't always work.

In fact, sometimes it doesn't work at all. There's been days when it has given me a solution, I've installed it, cleared the cache, refreshed the page, and nothing. Absolutely nothing changed. So I'd go back and say, "That didn't work," and it would come back with something like, "Ah, I see the problem." Well, apparently it didn't see the problem because the next solution didn't work either, and we'd go back and forth, sometimes for quite a while. 

But here's the interesting part. Most of the time, we eventually figured it out, and along the way, I started realizing that I could do things with my websites that I probably wouldn't even have attempted before. Not because I suddenly became a computer programmer, believe me, I didn't, but because I had something available to me that could help turn my ideas into working software, and I didn't have to settle for something that was almost what I wanted. 

I could keep working on it until it did exactly what I wanted. Well, within reason. There's a few times when I finally told it we'd spent enough time on something and it was time to move on, but that's part of the story too. 

So today I thought I'd tell you about why I decided to rebuild my business website, some of the things we've created, some of the problems we've encountered, and what I've learned from the whole experience. 

And I think there's something here that could be useful to just about anyone who runs a business . 

Now, before I get into all the things that ChatGPT and I have been doing, I should probably explain why I decided to rebuild my website in the first place. And it wasn't because there was anything particularly wrong with the old one. 

Actually, I was pretty proud of it. I rebuilt it myself about five or six years ago, I guess, using Drupal, and I'd even received compliments from other website designers about how good it looked. So it wasn't a matter of having an ugly website that needed fixing. The problem was entirely different. At the time, I had two business websites. 

One was Seagull Technologies, my primary business website, and the other was seagulltechnologiestraining.com. I'd created the training website with certain ideas in mind , but it never really developed the way I'd envisioned it. And after thinking about it for a while, I decided that maintaining two separate business websites didn't make a whole lot of sense. 

There were things on the training website that I wanted to keep, but I didn't really need an entirely separate website for them anymore. So I decided to consolidate the parts I wanted to retain into the main seagulltechnologies.com website. But I also wanted to do something more than simply move a few pages around. 

I wanted to rethink what the website was really about. You see, there's a philosophy I've come to believe in very strongly over the years, and it goes something like this: "At the end of your days, the only thing that really matters is how many people's lives you've improved along the way." Now you can agree or disagree with that philosophy, but it's become increasingly important to me. 

And when I started thinking about what I wanted Seagull Technologies to represent, I realized I'd accumulated quite a bit of knowledge and experience over the years in business, taxation, technology, education, and a variety of other things. And I started asking myself, how could I use these experiences and whatever talents I have to help other people? 

And that became the foundation for what I wanted the new website to accomplish. I didn't want it to just be another website advertising a collection of services. I wanted it to become a place where people could find informational, educational resources, and practical assistance that might actually make a difference in their lives. 

And that meant bringing together things that had previously been spread across two websites. Now, I should mention something about the last time I'd rebuilt seagulltechnologies.com. I think that'd been about, oh, five or six years ago. And I did it pretty much the traditional Drupal way. I searched for contributed modules that would do what I wanted. 

I installed them, I configured them, and I tried to get everything working together. I even learned a little inline CSS along the way. 

For those of you who don't know what CSS is, it's the code that controls how things look on a web page. And believe me, learning just enough CSS to make things look the way you want can be an adventure all by itself. 

I'd get something looking just right on my computer, only to discover that it didn't quite look as good on a phone or a tablet. Or I'd change the appearance of one thing and accidentally change something else, and then I'd have to figure out what I'd done, and eventually I got the whole website looking the way I wanted it. And as I mentioned, I was pretty pleased with the results. 

So, when I decided to rebuild it this time, I knew what I was getting into, or at least I thought I did, because this time I had something available to me that I hadn't had five or six years ago, an AI. 

Now, I have looked around at several AIs. Some of them were so technical that it was hard to follow along, and others just didn't give me the type of answers that I needed, and that's why I finally ended up with ChatGPT because it would sit and it would talk to me like it was an employee I had hired. 

I love to teach, but I love to learn also. And I learned a lot. I learned how to recognize what the errors meant when errors came up and how to fix them without having to ask how to fix them and things like that. Anyway, I started wondering whether artificial intelligence could help me accomplish some of the things that had been so difficult the last time around. 

And at first, I was thinking primarily about making the website development process easier. I had no idea that we'd eventually be creating custom tax planning calculators, online educational platforms, and all sorts of other functionality, but that's where the journey led Now, there's something I want to make clear. 

I didn't suddenly stop using Drupal modules. I still use quite a few of them, and some of them are excellent. Drupal has a tremendous amount of functionality available, and it would absolutely make no sense to start writing custom code to replace everything. 

But I began to approach things differently . Previously, if I wanted the website to do something, my first question was usually, "Is there a module available to do this?" And if there was, I'd try to make it work. Sometimes it did exactly what I wanted, other times not so much. But when I started working with an AI, I could begin with a different question: What exactly do I want this website to accomplish? 

Then we could figure out the best way to do it. Sometimes that meant using an existing module, sometimes it meant modifying the way a page was displayed, and sometimes it meant creating an entirely new custom module, which I'd never done before. That was the part that really opened my eyes. Because instead of adapting my business requirements to fit somebody else's software, I could develop software around my business requirements. 

That's a pretty significant difference. Of course, it didn't mean I could simply describe something and expect it to work perfectly. We still had to test things, we had to troubleshoot problems, and we still had to make decisions about whether something was worth the time and effort. But I was no longer limited to what I could find in Drupal's module directory. 

And at about the same time, I had a real business reason to put that new approach to the test. My first project that really opened my eyes involved a client who purchased a tax consultation through my website. He was, like, 63 years old and getting ready to retire. He had approximately, oh, a million and a half in a traditional IRA, and he was looking ahead to what would happen when he reached the age where he'd have to begin taking required minimum distributions. 

Now, for those who aren't familiar with the difference between a traditional IRA and a Roth IRA, let me explain something. With a traditional IRA, the money generally grows tax-deferred. But when you take it out, the taxable portion is usually subject to ordinary income tax. 

With a Roth IRA, qualified withdrawals are generally tax-free, and that can make a tremendous difference in retirement. Well, my client wanted to spend the next 10 years moving as much money as practical from his traditional IRA into a Roth IRA. Of course, there's a catch. 

When you convert money from a traditional to a Roth IRA, the taxable amount of the conversion is generally included in your income for that year. So you have to be very careful about how much you convert. And this particular client had a very specific objective. He wanted to convert as much as he reasonably could each year without moving into the thirty-two percent federal tax income bracket. 

I mean, when you jump from twenty-four to thirty-two, you're paying a lot more in taxes. His thinking was that by the time he reached the age where he expected to begin taking required minimum distributions, he'd have substantially reduced the amount remaining in his traditional IRA. That could reduce the taxable distributions he'd be required to take later and potentially lower his future income tax liability. 

Now, this wasn't the first time I'd helped a client with this type of planning. I'd done similar calculations by hand in previous years. And for a single year, that's certainly manageable . 

But this client was looking at a ten-year planning period. That meant evaluating the amount he could convert each year, considering his other income, the tax brackets, deductions, and other factors that could affect the calculation, and doing it again the following year, and the year after that, and the year after that. 

You get the idea. 

So I started thinking, "There's got to be a better way to do this." And that's when I turned to ChatGPT. I just explained the situation, what I wanted the calculator to accomplish. I wanted to be able to enter the client's income information, account for the relevant tax rules, determine how much additional income could fit within a particular tax bracket. 

And because this involved retirement planning, we also needed to consider how additional income could affect the taxation of Social Security benefits. Now, I don't want to give you the impression that the AI immediately produced a perfect calculator. It certainly did not. 

We went back and forth on the calculations, the inputs, and the way the results should be displayed for several days, and I had to pay close attention to the tax logic it was producing, even if it was accessing all of the manuals on the IRS website. 

Because this was a real client making real financial decisions. 

I couldn't just assume the results were correct because the computer produced them. But eventually, we developed a working tax planning calculator, and the best part was that now I had a tool I could use again, not just for this particular client, but for other clients facing similar decisions. Something that had previously required a considerable amount of manual calculation could now be handled much more efficiently, and that was when I really started appreciating the possibilities. 

Now, one thing I should mention is that this type of planning isn't simply a matter of converting as much possible every year. I mean, tax laws change, income changes, investment values change, and sometimes paying a little more tax in one year may or may not make sense when you consider the overall retirement picture. 

The calculator doesn't make those decisions for me. What it does is help me evaluate the numbers so I can spend more time thinking about the planning decisions themselves. And to me, that's a much better use of my time. 

Then about a month later, I had another client with a completely different problem. The client received a substantial amount of his income from investments, and unlike wages where an employer withholds federal taxes, much of this investment income had no withholding on it. That meant that he needed to make estimated tax payments during the year. Now, estimated taxes can be a real headache, and most people have a lot of problems with it. You have to estimate your income for the year, calculate the expected tax liability, account for any withholding or payments already made, and then determine how much additional tax needs to be paid. 

And if you don't pay enough or you don't pay it at the appropriate times through the year, you may end up having to pay an underpayment penalty. And so this client wanted to get these payments right, and I thought, "Well, here we go again." Another situation where I could do the calculations manually, but where a properly designed calculator could make the process considerably easier.  

So back I went to my AI. This time, we started developing a personal estimated tax calculator. And of course, it turned out to be more complicated than simply multiplying income by a tax rate. We had to account for different categories of income, deductions, credits, withholding, and the rules surrounding estimated tax payments.          

We also needed to think about how the results would be presented so they would actually be useful. And because tax figures change from year to year, you needed a way to update those figures without rewriting the entire program. And that may not sound like a particularly exciting feature, but if you're going to use a calculator year after year, it makes a tremendous amount of sense. 

We also worked on the reports because I wanted the information to be presented clearly, and I wanted the results to be printable. And that led to some of those little formatting issues that can take more time than you expect. I tend to be a little picky, which is probably why I spend twice as much time as most people have. 

But if you're gonna do it, do it right. Because something might look fine on the screen, and you print it, and suddenly the spacing isn't right, or something is positioned differently than what you expected. So back we go again. Another adjustment, another test. But eventually, we got it looking just the way I wanted it. 

We got the calculator working. And here's the part I particularly enjoyed. A couple of weeks later, another client came along with essentially the same problem. Well, at this time, I'd already had the calculator. I didn't have to start from scratch. I didn't have to spend time developing another spreadsheet or manually reporting all the calculations. 

I could use the tool we'd already built. And that really brought home the business value of what we were doing. We weren't simply saving time on one project. We were creating tools that could continue providing value whenever a similar situation came along. And because the calculators were integrated into my new website, I could make them available to authorized tax clients as part of the services I provide. 

That was becoming a pretty useful arrangement. 

Now, just when I was beginning to think we had this calculator business figured out, another client presented a different challenge. This one involved a complex trust. And if you've ever dealt with the income taxation of a complex trust, believe me, it can get complicated in a hurry. 

Well, the client had received a letter from the IRS. The estimated tax payments for the trust had been handled incorrectly, and now they were dealing with the consequences. Well, I certainly didn't want them going through that again, and I started thinking about whether we could develop a calculator that would help them evaluate the trust income tax liability and make better decisions about estimated payments. 

Now, trust taxation is a totally different animal from individual taxation, and for that reason, a lot of tax preparers don't even want anything to do with them. For one thing, trusts have very compressed income tax brackets. A trust can reach the highest federal income tax rate at a relatively low level of taxable income. 

And then there are the rules involving income distributions to beneficiaries. Depending on the circumstances, distributing income can affect how much tax is paid by the trust and how much is reported to the beneficiaries. Then there's something called distributable net income or DNI, which plays an important role in determining how those distributions are treated. 

Then you have different types of income deductions, capital gains, and net investment income tax, and there are special timing rules that can affect certain distributions. Now, I'm not going to turn this podcast into a course on trust taxation, but I think you can appreciate that this wasn't going to be a simple calculator. 

And once again, I had a very specific reason for wanting to build it. I wanted to help a client who already had encountered problems with the IRS avoid repeating those problems. So, ChatGPT and I started working through the requirements. We developed a custom module that could accept the relevant income and deduction information, evaluate distributions, and estimate the resulting tax liability. 

We also included a way to examine what could happen if an additional distribution was made, And that's particularly useful in trust planning because the tax consequences of retaining income in the trust can be quite different from those distributing it to the beneficiaries. 

We also created administrative settings so that tax figures could be updated for future years, and we had to make sure the calculator was available only to authorized clients. 
So I didn't start by asking the AI to come up with some interesting software I could put on my website. I started with problems I was already trying to solve for clients, and that's what really made the whole process so valuable because once we built a calculator to address a particular problem, I could use it again. 

Of course, I still had to understand the tax rules, and after a quarter of a century of doing small business taxes and trust taxes and all the rest, I had that experience. I still had to review the calculation, and I still had to make the business decisions. But I didn't have to become a professional software developer to create the tools, and that, to me, was the real breakthrough. 

I wasn't looking for ways to use AI. I was using AI to solve problems I already had . And once I started thinking about it that way, a whole lot of other possibilities began opening up. Now while all of this was going on, there was another project I wanted to bring into the new Seagull Technologies site, my business course. 

I developed a course called Mastering Business Basics, and the purpose was to help people understand some of the fundamental things they need to know about starting and operating a business. Because one of the things I've observed over the years is that people sometimes go into business knowing a great deal about the product or service they want to provide, but not necessarily knowing very much about actually operating a business, and those are two completely different things. 

Someone might be an excellent carpenter, mechanic, consultant, or whatever, but that doesn't automatically mean they understand financial management, taxes, legal structures, or marketing, and those things matter. So I developed educational material to help people build a better foundation. The courses included videos, written transcripts, and additional reference materials. 

What I needed was a way to deliver that material through my website. Now there's plenty of online course platforms available, and I'm sure some of them are very good. But I already had the content, I already had the website, and I wanted the course to be part of Seagull Technologies rather than sending people off to an entirely different system. 

So once again, I started talking with my AI about what I wanted, and this became another one of those projects that grew as we worked on it. First, we needed a way to organize the lessons. Then we needed a classroom where students could watch the videos and read the transcripts or listen to the audio. 

Then we needed to keep track of their progress. And of course, we needed to make sure that only people who had purchased the course could access the material, so we had to develop an enrollment system, and we had to connect that enrollment system to the payment process. Then I started thinking about assessments because I didn't want students simply watching a collection of videos, listening to audios, or reading transcripts and then moving on. 

I wanted them to have a way to check whether they understood the material. So we developed a question bank and a system for recording quiz attempts, and we needed administrative tools so I could review and manage the questions. Now, the course itself took quite a bit of work. We had to develop the class ones. 

We had to organize the lessons. We had to make sure the progress tracking worked. We had to handle enrollment and permissions. But the result was a custom educational platform designed around the course I'd already developed, and that's exactly what I wanted. Once again, I wasn't trying to make my educational material fit somebody else's software. 

We were developing software to fit the educational material. 

Now, there's another part of this story that I think is worth mentioning, because building a website that provides information is one thing, but I also wanted the website to handle some of the ordinary business processes that go along with offering services. For example, I offer tax consultations and business mentoring, and I wanted someone to be able to visit the website, purchase a consultation, provide some information about what they wanted to discuss, and schedule an appointment. 

Well, that sounds simple enough, but there are several things that have to happen. First, the person needs to pay, then the website needs to recognize that payment was successful, then the person needs to provide the information I need before the consultation, and finally, they need to schedule a time. So for payment processing, we use Stripe. 

For scheduling, we use cal.com, and for collecting information, we use the forms within Drupal. Now, this is really a good example of something I think is important. Just because AI can help write custom software doesn't mean you should write everything yourself. I had absolutely no interest in creating my own credit card processing system. 

Why would I? I mean, Stripe is one of the best in the industry, and they already provide that service. And I didn't need to build an entire appointment scheduling platform when an established service could handle it and integrate it automatically with all the systems I have on my computer. What I needed was a way to connect those services to my business process, and that's what we worked on. 

So we developed the payment flow, the confirmation process, the intake forms, and the scheduling pages. We also had to think what would happen if somebody tried to access a page they weren't supposed to reach. For example, you don't want someone to bypass the payment process simply by typing the address of a scheduling page. 

So we worked on controlling access and verifying the payment processes. While this is not particularly glamorous, these are a lot of things I probably would have never considered before or known how to do properly. But it's the kind of work that makes a difference between having a website that looks nice and having a website that actually supports your business. 

And it was another area where I appreciated being able to explain what I wanted rather than trying to assemble an assortment of unrelated software Now, I don't want to give you the impression that everything we worked on was some enormous software development project. 

A surprising amount of our time, in fact a lot of our time, was spent on little things, getting things just right. And sometimes those little things turned out to be the most frustrating of all. For example, we worked on the appearance of the buttons on the site, the spacing between the sections on a page, the way the menu operated, the position of all the navigation elements the way I wanted them, and whether a horizontal line looks better above or below a publication date. 

Yes, we got that picky . And you know what? Those details matter. Maybe not individually, but when you put them all together, they affect how professional and comfortable a website feels to the person using it. One example involved the navigation at the bottom of some of our content pages. 
I wanted it centered. That sounds like a five-minute job, and sometimes it is. But something as simple as centering an element can require finding the right CSS instructions. We had situations where ChatGPT provided a CSS change, I installed it, and nothing happened. 

Then we'd have to inspect the HTML and discover that the element had a different class. We'd have to change the CSS, clear the cache, reload the page, and finally, we found the right one, and it was exactly where I wanted it . 

Now, I realize that doesn't sound like a major accomplishment, but when you've spent an hour trying to get something to move a few pixels, you can get pretty excited when it finally does. 

And I think those small victories are part of what made the process enjoyable also. Well, most of the time. There were also times when the AI and I went down what I can only describe as a rabbit hole. We'd start investigating a problem, it would suggest something, I'd try it, it wouldn't work, then we'd try something else, and before long, we'd be examining files and settings that I hadn't even known existed before. 

Sometimes that was necessary. Other times, we'd eventually discover that the problem was something that was fairly simple, and every once in a while, I'd finally say, "You know what? We've spent enough time on this. I don't need those few extra pixels right there. ". 

Because there's another lesson I've learned. Just because you can keep working on a problem doesn't mean you should. Sometimes you have to decide whether the problem is worth any more of your time, and that's a business decision. 

I've learned that I prefer working through technical problems one step at a time, and there's a reason for that. When you're dealing with a complicated system like Drupal, making several changes at once can create a real mess. Suppose it gives me five different things to change. 

I make all five changes, then something goes wrong. Which change caused the problem? I don't know. Then I've got another problem to solve. So we've increasingly worked with a simpler approach. Give me one step. I'll make the change. I'll tell you what happened. Then we'll decide what to do next.  

Sometimes that means copying a command to a server's terminal, sometimes it means adding configuration files, sometimes it means changing a setting in Drupal, and sometimes it means simply checking whether a page works. But by doing one thing at a time, we can usually identify where something went wrong. 

And if a change doesn't work, we can deal with that before moving on. I've also learned that I don't need to understand every line of code to participate meaningfully in the process, but I do need to understand what the code is supposed to accomplish . 

And I need to know enough about my website to recognize when something isn't behaving correctly. And that combination has worked pretty well for us. 

we've also spent time dealing with security settings. Now, security is one of those subjects I don't believe in taking unnecessary chances. In this world of hackers, you can't. 

So we worked on access permissions, uh, whether certain pages were available, people shouldn't be able to see them, examined server settings, worked on some restrictions that control which outside services can communicate with the website. That last part can get surprisingly complicated, and I'd never done any of that before because I didn't understand how to do it. 

And if you make your security settings too restrictive, you can accidentally block something that the website actually needs. So you have to find the right balance, and it's been a great help. And again, this is where I think it's important to understand what you're trying to accomplish rather than simply copying code because the A I suggested it. 

I mean, artificial intelligence can be very helpful, but it is not a substitute for your judgment. And there's certain things it can't do. it didn't decide what my business needed. It didn't come up with a philosophy about helping people . It didn't determine how I want to work with my tax clients, and it didn't decide what subjects I should teach in my business course. 

Those decisions came from my own experience and my understanding of the business. What it did help me do is turn those ideas into working systems , and sometimes it helped me think through alternatives I hadn't even considered. But I still had to make the final decisions. I had to evaluate the results, and I had to determine whether the final product actually accomplished what I wanted it to. 

I think that's an important distinction, because there's a lot of discussion about artificial intelligence replacing people, and I certainly understand why people are concerned about that . But my experience has been much more about extending what I can accomplish. I didn't suddenly become a professional programmer. 

I didn't need to . I had a tool that could help me with the programming , just as if I had hired an employee, and I had enough knowledge of my business and my website to guide the process. That combination has been remarkably productive. And there's another thing I've noticed. The better I can explain what I want, the better the results tend to be. 

When I explained the client's situation, the calculations I needed, and how I wanted to use the results, we had something specific to work towards. And I think that's a lesson that applies to almost any use of artificial intelligence. You don't necessarily have to know how to do the technical work, but you do need to know what you're trying to accomplish. 

When I look back at the last several months, I'm actually a little surprised by how much we've done. We started with a decision to consolidate two business websites into something much bigger. We've redesigned and reorganized the company around the things I wanted to accomplish. We developed custom tax planning calculators based on real client needs. 

We created online classrooms for my business course. We integrated payments, intake forms, appointment scheduling, worked through all sorts of appearance and navigation issues, and we dealt with the inevitable technical problems that come with maintaining a website. 

If you're thinking about using AI to help you with your own business, here's a few things I suggest based on my experience. 

First, don't start by asking what AI can do. Start by thinking about what you want to accomplish. Maybe there's a repetitive task that's taking too much of your time. Maybe there's information you need to organize. Maybe there's a calculation you performed regularly, or maybe there's something you've wanted your website to do, but haven't been able to find the right software. Start with the problem. 

Second, don't assume the first answer is going to be the best answer. Sometimes it will be, and sometimes you'll need to explain more, provide more information, or try a different approach. 

Third, don't be afraid to tell the AI that something isn't working. Believe me, I've done that plenty of times. And if a suggestion doesn't make sense, question it. You don't have to accept an answer simply because it came from a computer. Sometimes they get so technical they don't see the simpler solution. 

Fourth, always test the results, especially if you're dealing with money, taxes, customer information, or anything else where an error could have serious consequences. 

And finally, don't forget that your time also has value. Just because an AI can help you pursue a solution doesn't mean every solution is worth pursuing.  

Sometimes you need to decide that something is good enough, and sometimes you need to decide that it's time to move on . I've had to remind ChatGPT of that on more than one occasion. 

So was it worth it? Absolutely . 

Not because every project was easy, and certainly not because everything worked the first time, and because I now have capabilities on my website I probably wouldn't have attempted to develop otherwise. 

And more importantly, those capabilities are connected to what I wanted the business to accomplish in the first place. Remember that philosophy I mentioned earlier? "At the end of your days, the only thing that really matters is how many people's lives you've improved along the way." That's what the whole project was supposed to be about. 

The tax calculators weren't developed simply because I thought it would be an interesting thing to write some software. They came about because clients had problems I wanted to help them with. The business course wasn't developed because I wanted to experiment with an online classroom . 

It came about because I wanted to share information that could help people become more successful business people. And the website itself wasn't rebuilt because I was unhappy with the way it looked.

It was rebuilt because I wanted to bring together the different ways that I could use my knowledge and experience to help others. Artificial intelligence simply gave me the new tools to accomplish those things. 

And that's what I find so exciting about it. 

Not the technology for its own sake, but what that technology makes possible. But building it was only half the battle. Now you'd think that after all that work I'd be ready to sit back and enjoy the results. Well, not quite.

Because there's another problem. You can build the most useful website in the world. You can fill it with educational material. You can create calculators, courses, and all sorts of helpful resources, but if nobody knows the website exists, what good does it do? 

And that's where the next part of our adventure begins. We started looking at more closely at how people actually find websites. We looked at Google, we looked at Bing, we started examining search results and analytics And we've been asking some interesting questions about how artificial intelligence might change the way people discover information on the internet. 

So we also started making changes to the way we publish content. and that is what the next episode is going to be all about. I think you'll find it interesting. Until then. 

Announcer: You have been listening to the Mastering Business Basics podcast with your host, Roger Pearson. For more information about all of the business education options that are available, we invite you to visit seagulltechnologies.com and continue your journey.