During a recent audit, the investor asked us whether we felt the product had a moat from a technical perspective.

Much like the moat of a medieval castle, the technical moat is a defensive barrier, making it harder for competitors to swoop in. Typical examples are proprietary tech or IP. If you have that, you have a competitive advantage.

In these days of the AI boom, it's often a polite way of asking, "Is this just a ChatGPT wrapper?'

Many products built today claim to be Artificial Intelligence, but just push some prompts to OpenAI. From an investor's perspective, that's a risk. Such a company is at the mercy of its LLM vendor. Changes to the pricing or T&C can hurt the client— or worse. Where a lot of companies with IP get acquired by the big ones, nothing is stopping a Microsoft team from flat-out copying a successful ChatGPT wrapper as a Word feature and killing the business.

It's understandable that investors consider such a risk.

But there's another factor at play. The alternative to OpenAI is often a technical risk, too. Imagine a team spending a year building their own LLM rather than going to market on top of OpenAI. The cost of training, tweaking, DevOps, maintenance, infrastructure, and research would be off the charts. Hosting an open-source LLM mitigates some of the risks but comes with operational complexity.

After that year, that company would end up with something considerably worse than OpenAI. They would have a moat, but nobody would care about storming their castle.

I feel the nuance is whether the product is AI or has AI — Whether your customers consider you a technology vendor or a SaaS.

If you're selling a technology, you need a deep moat. You need to build your own tech and IP. But if your product is tailor-made to solve the expensive problems of a clear target audience, they won't care what's behind the curtain.