Yesterday I posted about how a friend built an AI Chief of Staff for himself (see here). Another thing we discussed was application software and its vulnerability to AI. My friend is technical by training and understands the underpinnings of both technologies better than I do, so I was curious about his take.

He didn’t hold back. He predicts that in two or three years, many paid software applications will begin to be replaced by free software written by AI. As AI improves, he thinks it will empower the normal person to create software that solves their specific problem extremely well (better than current paid application software products). His big point is around context. Right now, for AI to correctly build what you want, you must give it the right context and instructions. If you’re not technical and can’t instruct AI on what tools to use and how to use them to build what you want using the right approach, it will build something bad. But if you’re like him, your instructions are great and it will write the software you want in 10 minutes.

Think of it in terms of building a house. Right now, AI is the general contractor. It can build whatever you want. But to understand what to build, it needs to work with plans drawn up by an architect. Imagine just giving a general contractor a description of what your home should look like. Would it build what you envisioned? No chance.

My friend’s point is that in two or three years, you won’t need a background like his or the ability to instruct AI. You won’t need the architect. AI will be so good that it will have enough context to build what you want straight out of the gate. AI will become the general contractor and the architect all in one. When that happens, it will be hard for application software companies to retain and attract new customers.

This was an interesting take. It’s definitely something I want to digest and think more about. I appreciate my friend sharing his take; I always learn a ton when we chat.