Working With AI, Not Just Using It: Building a Dive Business in Mexico

I started using AI to build a website because I thought it would be quicker. Several months later, I was still working on it, which wasn't quite what I'd planned.

Somewhere along the way, though, I'd developed a new business brand, built a website competing in Google searches with dive centres that have been established for decades, learnt far more about modern SEO than I'd intended, and started using AI for all sorts of things I hadn't originally considered. There was also an underwater video that went viral, made the Mexican news and eventually led to a scientific collaboration, but more on that later. What began as a straightforward website project had turned into something rather bigger.

A Different Sort of Colleague

I've spent much of my career managing people, working with teams and developing systems. Whether dealing with operational problems, business decisions or new projects, I've generally preferred to work collaboratively: talk through the problem, throw around some ideas, challenge each other, make a decision and get on with it. It turns out that's pretty much how I use AI.

I don't generally sit down and write an elaborate prompt telling it exactly what to produce. More often, I start with an idea and we have a conversation. Sometimes I know exactly what I want, while other times I'm still trying to work that out. We discuss options, challenge suggestions, explore alternatives and gradually arrive at something useful.

Of course, AI doesn't actually have opinions or get offended when I tell it an idea is rubbish, which is probably just as well. It can also be incredibly frustrating, occasionally making a perfectly sensible suggestion before contradicting itself five minutes later, spending ages explaining something I've already understood, or seeming determined to solve a problem I didn't actually have. Anyone who's worked with me will probably recognise my response to that.

When it works well, though, it's remarkably effective. I've found that many of the skills I've developed working with people are just as useful when working with AI, particularly the ability to ask questions, challenge ideas and provide direction. The important difference is that AI doesn't take responsibility for the result. That's still my job.

Building the Brand Before the Website

Before getting too involved in building the website, we needed to establish what Mahahual Underwater Explorers was going to represent. Mahahual already has plenty of dive operators, and I didn't see much point in creating another website that looked and sounded like all the others.

My experience is primarily in professional diver training, technical diving and underwater exploration. I wanted the business to reflect that background while remaining approachable for someone considering their very first dive. We spent time working through the brand identity, visual presentation, messaging and how the different parts of the business should fit together. Some ideas looked good initially but didn't quite represent what I wanted, while others developed through discussion and several attempts.

By the time we started building the Mahahual Underwater Explorers website, we had a much clearer picture of what we were trying to achieve. The branding wasn't simply about a logo and some colours; it shaped the way we presented the business, described our services and communicated with potential customers. Then came the supposedly quick job of putting it all online.

The Mahahual Underwater Explorers homepage, developed through collaboration with AI

Old Skills, New Rules

I'd built websites before and knew about SEO, so I wasn't starting from scratch. The problem was that much of my knowledge was out of date. Google had moved on, the tools had changed, and there was considerably more to think about than I remembered. I found myself spending hours looking at search results, comparing other dive operators' websites, working through Google Search Console and trying to understand why one page ranked well while another barely appeared.

AI was useful here, not simply because it could suggest changes, but because I could ask why. Why should we structure a page this way? Why is a competitor ranking above us? Does that keyword actually matter? What happens if we change something that's already performing well? And importantly, how do we know whether a recommendation is any good?

We made changes, watched the results, sometimes changed things back and occasionally decided the best thing to do was leave a page alone. It certainly wasn't instant, but today the website performs strongly for a range of diving-related searches, sometimes appearing ahead of businesses that have been operating in the region for decades. That's rather satisfying for something I originally wanted to knock together quickly.

The Advantage of Actually Knowing Your Subject

One of the biggest lessons came from something that had very little to do with technology: I know diving. After years of teaching, training instructors, technical diving and working in the industry, I have a pretty good understanding of what divers want to know. As we reviewed other websites, we noticed that many spent a lot of time telling visitors how wonderful an experience would be without clearly explaining what they were actually getting.

How much does it cost? What's included? How long does it take? What equipment do I need? What happens if I'm arriving on a cruise ship? What experience do I need for a particular course? These aren't difficult questions, but they're often exactly what people are looking for when deciding whether to book.

So we concentrated on answering them, developing detailed content about subjects I know particularly well, including technical diving, underwater scooters, night diving and diver training. Rather than simply repeating generic descriptions, we could explain things properly, include practical detail and answer questions from experience.

As we developed more useful, detailed content across the website, our Google rankings improved. I wouldn't pretend to know exactly how Google's algorithms made those decisions, as there are plenty of factors involved, but the experience reinforced something that makes sense to me: there's a real difference between generating lots of content and publishing information that genuinely helps people.

AI is exceptionally good at helping organise and communicate knowledge, but if you don't understand the subject well enough to recognise what's accurate, useful or missing, you can end up with very polished nonsense. I suspect that's going to become an increasingly important distinction as more businesses use AI to produce their online content.

Beyond the Website

Once the website was established, the projects started multiplying. Google Ads introduced another learning curve, and we began using AI to help develop practical systems for running the business.

One of the more interesting projects has been building a connected business management system using Zoho CRM, Make and Airtable. It links customer enquiries and bookings with equipment management, allocation, servicing and maintenance records, reducing the need to manage everything separately. The system and its automations are now operational, with the remaining work primarily involving populating the equipment database with all the individual assets.

I've spent years developing systems and procedures, usually working with other people who brought different technical expertise to the project. This time, I've been doing much of the practical implementation myself, with AI helping me work through database structures, integrations and automation. It's a fairly detailed setup for a small dive business, but I can see the value in getting it right from the beginning.

Our connected business management system, integrating Zoho CRM, Make and Airtable for bookings, equipment management and maintenance.

Not everything worked first time. There were plenty of changes, troubleshooting sessions and moments when I wondered whether simply using a spreadsheet would have been easier, but seeing the different applications now working together has been particularly satisfying.

We've also been developing our social media presence, sharing diving experiences through Instagram and YouTube. That's where something rather unexpected happened.

During a dive off Mahahual, I filmed a sea turtle releasing her eggs directly into the ocean rather than laying them on a beach. We shared the footage on Instagram, where two posts featuring the same video attracted more than 800,000 views combined. The video took on a life of its own, being picked up by hundreds of social media accounts and featured in the Mexican news, where it became part of a wider discussion about the possible impact of sargassum on turtle nesting.

But the most surprising development was the scientific interest. The footage caught the attention of researchers and led directly to a collaboration on a research paper now being considered for publication. Not bad for an underwater video from a small dive business in Mexico.

YouTube Video showing the sea turtle releasing eggs underwater.

So, Has AI Saved Me Time?

Yes and no. There are things I've accomplished in minutes that might previously have taken hours, but there are also things I've spent entire evenings working through that should have been straightforward. AI can be incredibly useful, but it can also be confidently wrong, misunderstand instructions or offer contradictory advice. I've learnt that treating its answers as recommendations rather than facts is essential, particularly when dealing with business decisions or technical information.

Looking back, it's rather amusing that I started this because I wanted to save time. Instead, I've spent months experimenting, learning and occasionally arguing with a machine. But we've built a brand and a website that can hold its own against businesses that have been around for decades and opened up possibilities I hadn't even considered when we started.

And it's very much a work in progress. If you're curious to see how it's developing, have a look around the Mahahual Underwater Explorers website, follow our underwater adventures on Instagram or explore our videos on YouTube.

It turns out AI wasn't quite the shortcut I expected. But it's become a pretty interesting way to work.