My friend, William Neil, recently wrote an app. Rather, he convinced AI to write an app for him. And he recorded himself doing it!

If you’re technical, you might enjoy watching it. Or, if you find recordings of machines generating machine-y looking things soothing like ASMR.

William chose one of my collaboration workouts as the subject of his app. I host a monthly series of workouts called Staying Strategic as part of my Collaboration Gym. The first two workouts is something called Question Dumping — literally dumping whatever questions or top-of-mind for you at the moment, then doing a little organizing and answering. Participants do their work in Google Docs, and I prompt and coach them along the way.

William did the program earlier this year, and he asked me if I had ever thought of creating an app version of my workouts. As a matter of fact, I’ve often thought about this. So, he built one!

We had an hour-long conversation where I walked him through how I thought it should work, then he spent at least another hour refining the requirements using ChatGPT as a kind of secretary while he went on a walk (as he describes in the video). Building / testing / debugging an iteration of the app took about an hour total.

I haven’t done any serious coding in decades. I also haven’t played with coding using AI, although I’ve had some interesting conversations with friends, and I’ve been following musings from Jon Udell, Les Orchard, and Harper Reed. (I also enjoy Mike Hoye’s general commentary and critiques.) So it was fun to have William walk me through the process, and I’m so glad he recorded this and shared it with the wider world.

It’s a simple, non-production quality app. I think an experienced engineer could also build it in about an hour, with other potential benefits, such as higher code quality. It also would have cost at least an order of magnitude more. Maybe the benefits outweigh the costs in the long run. I wonder about things like maintainability and how easy it is for LLMs to refactor a codebase with minimal human intervention. And, I’m also conscious of the Bitter Lesson.

Now about that app…

For many years now, even before GPT-3, my friend, James Cham, has talked about the potential of AI replacing overpriced strategy consultants. If you look at the mechanics of what most traditional strategy consultants do, this is a no-brainer. But the dirty little secret of Big Consulting is that they’re rarely hired for said mechanics.

So I don’t think that AI will replace Big Consulting (although it may drive the cost down). But I do think that many groups legitimately want to align their people around good strategy. This is why I created Staying Strategic, which is an evolution of something I’ve been tinkering with for my whole career. And I think that AI could not only help scale workouts like Staying Strategic, but it could enable new and interesting possibilities. The fact that AI could help create the tools that would enable AI to help us be more strategic is just meta icing on the cake.