I always wanted to have my own personal CRM system. Not an enterprise CRM, but something more shiny and fun to use instead. And I have never found something that was good enough to keep using after a few weeks.
As a benchmark, Things 3 has always been one of my favorite MacOS apps. The pastel colors and extensive use of spacing makes it enjoyable to use. It is an app that always makes me want to get back into checking of items of my to-do list. So would I be able to build my own CRM system that I have in my head, in a way that it is fun to use so that this time it sticks?
This will be an experiment that is spread over a longer period of time, as it would be the first app that I want to build that is more than just a simple thing to fix a small problem. Things like the earlier X bookmark tool from last week.
In my days as the founder of an app development company I often struggled explaining what it was I had in mind for an app we were building. And then after it was completed it was never completely what I had in mind. This had to do with the limits in communication and a budget that was depleted before the iterations were done.
The traditional flow I always went through was something like: draft a rough idea, create wireframes, back-and-forth with a designer. Then a developer builds it, I see it and realize it’s not quite right, more rounds of revision. Weeks or months before something real saw the light of day.
Would the extensive use of AI tools make it possible for me to build something from scratch that is exactly what I have in mind? And would I be able to make it into a version that I will keep using over time?
I want to figure out whether the dynamic between the vision-holder and builder has changed, and if so, in what way.
I started with a UI design that shows the features in its full design glory. This is different from how it is done in the traditional way. I skipped the wireframe phase we normally did. With AI there is no meaningful difference in the time it takes to create wireframes or the actual design. Way easier to grasp the flow this way.
And after a few short iterations with Claude Cowork working in Figma, I already got something that was looking the way I had in mind. Really cool to work this way. Here are some of the designs we made together.
Key Insight:
The costs of translating an idea into a design into a working app have dropped to almost zero. We don’t know what this means. It’s not the question whether AI tools are good enough, it is testing if we can remove the translation layer. Becoming the vision holder and the builder. Can something in my head finally make it to the screen fully intact?