The Promise of AI (Part V) - Be your own customer

If you wait for the customer to tell you what to do, you're too late, so we were our own customer

— Yvon Chouinard (source)

Now, thanks to AI, you can just build things. When deciding what to build, first look inward.

Without realizing it, I've been on the journey of being the first customer for a few months. Working independently, I took a hard look at my effectiveness and decided I needed to invest in doing more deep work. If I wanted to improve my deep work performance, I had to measure it.

Measuring deep work means using some form of a timer. To do this, I looked at a variety of options. Many were enterprise-grade and too complicated for what I was after. Other options focused on being "automated"; tracking what programs you use on your computer. The issue with this is two-fold: not all work worth tracking is done on your computer and an individual website means something in one context and something different in another. YouTube can be about using Claude Code in one context and watching English Premier League highlights in another. I wanted something simple that worked.

The best option I found was buying a timer from Korea with an accompanying app. Over time, I became increasingly frustrated by the buggy software, such as having to sign in what felt like every hour. As Rich Barton has noted, many businesses are born out of frustration.

In this case, however, prior to the advancements in Claude Code, I would have accepted the buggy software as the best I could do in that situation. Maybe after a few months I would have done another scan of the available options. After Claude Code helped me build jeremycowcher.com, I realized that maybe I didn't have to accept this anymore. I could build a deep work timer.

I called this idea Elementary Timer and became its first customer. Prompting Claude Code in VS Code, the timer has been far from a one-shot effort. I've had several hundred back and forths with Claude Code, covering issues ranging from syncing across mobile and desktop devices, refining the timer's shape and color, expanding the options for the tv shows / movie easter eggs, incorporating the ability to add offline sessions manually and breaking down how users spend their time.

The confidence that Elementary Timer gave me meant I could start on IdeaKache. This is so early I'm hesitant to mention it; the idea of IdeaKache is to store all the quotes, concepts and book notes I've captured in Dropbox over the last 10 years. It's data I'd captured purely for my own interests, with no plans to do anything with it. Prior to these AI advancements it would have likely stayed in my Dropbox. Now, with AI, that data might actually become something.

Are these two ideas any good? Maybe, maybe not. I do know that there are very few perfect ideas and if you had one, you'd already know that. Execution matters and to execute at a high level, especially early on, you need to build something you're interested in, you need to be the first customer.

The follow-through, the execution, is a thousand times more important than a "great idea." In fact, if the execution is perfect, it sometimes barely matters what the idea is

— Felix Dennis, How To Get Rich