This is an early version of a problem I still work on today: how do you remove friction from a process when the tools were designed for a person clicking through screens?
Why I made this
The workflow crossed application boundaries and depended on graphical interfaces. There was no simple single-system feature that turned the whole sequence into one action.
Repeating the same clicks manually would consume time and make the process fragile. The useful intervention was to teach a visual automation tool how to perform the sequence consistently.
What I did
I used Sikuli’s visual scripting approach to connect the Coda and Firefox steps, then recorded the construction process at high speed so viewers could see the script take shape without sitting through every pause.
The format preserves both the reasoning and the result. It shows that the automation was built against a real workflow rather than presented as a detached tool demo.
A note on the archive
Sikuli, Coda, Firefox, and desktop automation practices have changed since 2010, so don’t use this as a current implementation guide.
What I still like about it is how recognizable the problem is. Long before today’s low-code and AI tools, I was already looking at repetitive work and wondering how much of it I could make the computer do for me.