I don’t believe in defensive business moats. Moats are obviously important to a lot of strategists—Paul Ford said so last week. I’m just skeptical of their durability or reliability. It’s me, not you, dear reader. We can attribute my skepticism to a few factors:

• I’m Lebanese. A quick glance at Lebanon’s history over the last 50 years will tell you one thing: We really suck at moats. 
• I ran two agencies. (And Aboard is increasingly an AI digital transformation consultancy, so I’m now running a third.) An agency or consulting firm is arguably one of the most perpetually vulnerable businesses in existence. It is inherently moat-less. You can try to be as thorough and diversified and cost-efficient as humanly possible, but once you get that bad phone call, everything changes.
• I’m very paranoid. For better or worse. As a business leader, I spend a lot of time thinking about how the business will inevitably shit the bed. Even when good things happen—especially when good things happen—I assume the worst. It sort of works (except for the bad sleep and acid reflux). 

You’d think an ironclad contract would be the ultimate moat. Everyone in the consulting industry believes that. But we’ve dealt with some of the biggest companies in the world. When they want out, you have to let them out, contract be damned.

Welcome Aboard LLM Airways

So you’ve got your slick startup, and you’re surrounded by a quote-unquote “moat” that you’re building on top of one of the AI models. You’re going to make all sorts of things better, faster, and cheaper in sector X or industry Y. Insurance. Law. Marketing. Health. Everyone’s got it figured out. 

Your AI-powered solution is not really important here. If you raised a bunch of money, put out a product, and started generating revenue, the moat test will inevitably get thrown at you: 

What is inherently special about your strategy and product such that it is difficult for competitors to eat away at your market and maybe even poach your customers?

You would have been asked that question 20 years ago, too—but today is different. A moat implies others may attack your castle. In today’s “everything AI all the time” tech universe, the moat—and the castle, for that matter—is built on top of your potential competition.

If you’re wading into the AI space, you have to pass a very particular, much more terrifying test. Let’s ask a different question: 

Can [pick your LLM provider] that is powering your software credibly do what my product does?

Spoiler alert: The answer is probably yes. Claude or Gemini (or any other major LLM) are really good at parsing contracts and reading medical reports. And they’ll get better. 

Now couple those capabilities with the fact that:

• These LLMs could revise their terms of service and cut you off;
• AI providers carry outlandish valuations and they will inevitably swim upstream to get more revenue and;
• LLMs may not be doing a great job at what you specifically do right now, but they’re getting better every day.

If OpenAI needs revenue—and there is a lot of revenue in legal services—then they’ll be able to build your product and deliver it more cheaply and efficiently than you. They’re already doing it.

So is it game over? Nah. Here’s what we believe at Aboard, where we’re happily working with big clients in insurance and healthcare (literally the two most regulated industries on earth) to help them get incredible value out of AI (which is the least regulated):

1. User experience matters more than ever. The AI breakthrough was an undesigned experience. You’re chatting with a robot. As we introduce new products and offerings, we have to make them approachable and easy to understand. That still requires good, thoughtful design and user testing. Remember, people are too busy to thoroughly describe that solution that’s going to be great for them.
2. Learn your verticals. This has nothing to do with building. The more you understand the industry you want to be in, the bigger an advantage you can build. This isn’t just about “building the right app.” It’s about understanding the cultural, regulatory and even social limitations and ground rules for how a particular industry works. The status quo is notoriously resistant to change. You have to learn and speak their language to make any headway. 
3. The what is greater than the how. The execution part is getting rapidly commoditized. You can create just about anything now and the output is solid and fairly reliable. But what should you build? This requires patience, conversations, and real problem-solving. Creative problem-solving. This is the strategic part—good strategic thinking does not happen in a prompt box.
4. Relationships are far more durable than transactions. AI is a good talker but a lousy listener. It’s an even worse observer. That’s because it has no brain or memory. Everyone is anxious these days—and calmly building trust is more important than ever. You want to disrupt healthcare? Get ready to navigate the Byzantine world of healthcare norms and regulations. Let an LLM help you if you want, but ultimately, you’ll need to look someone in the eye and say, “Yes, it’s HIPAA-compliant.”

You don’t need to squint to see that all of these suggestions are about people and relationships. We’re seeing this first-hand at Aboard. The sales cycle isn’t shorter. In many cases, it’s longer. Are people the moat in the age of AI? Yes. Skills may come and go—most of us can’t do our own plumbing, and most stock trades don’t involve humans these days. 

As we go into organizations and see thirty years of calcified processes and brittle tools, the appetite for radical change is very low. You could never parachute in a killer app into an enterprise, and AI isn’t going to change that.

That said, change is coming. Paul sees it coming right now—in some ways he’s more paranoid than I am. But it’s definitely happening. In a world where anyone can spin up capability on demand, advantage comes from context, judgment, and relationships that compound over time. I expect to figure it out, and I expect you will, too.