Just a cool hack, or a good idea worth investing in?

Whether we're entrepreneurs or working at Big Corp, trying to leverage AI features in new or existing products, we need to need to tackle this question sooner or later.

"The AI Prof" Ethan Mollick of Wharton tackled this in a short LinkedIN post an hour ago: "Want to launch a startup that <monitors photos for problems, and if it finds any, alerts security/maintenance/emergency services>? No point, it is a native capability of GPT-4, and as soon as OpenAI implements agents, it will be easy to ask the system to contact you if it sees a problem. Hard to build a moat." [I'll link to this in the comments.]

Indeed, that's the crux of many product questions these days:

How can we use this AI stuff to build a value prop, with a moat that the giants can't crush in 2 months of dev time?

In the middle ages folks built big water-filled moats (sometimes stocked with alligators) around their castles to keep opposing armies from walking right up to the walls and scaling them with ladders or siege engines. As you know, a "moat" in competitive strategy describes our ability to create (and, importantly, maintain over the longer term), advantages over our rivals.

I have a few ideas of ways to build moats, but 250 words (the space of these "atomic essays") isn't a long time. Will write another one tomorrow.

Until then - what do you think?