Kanban Build

the times they are a-changin'

Kanban Build

Build & Repo here

Recently cranked out a fully functional local storage kanban board with Next, Zustand, a drag-n-drop library, and tailwind. The perks? The catch? This took less than a day, I estimated 4.5 - 5 hours with additional tweaks before deployment. Emulating the design system of their site (fonts and tailwind) was the logical starting point. Branding happens throughout development. Mimicking and improving upon existing designs can help ensure UX and UI functionality rhyme with older siblings in a growing brand's technological family tree.

Tasking GPT with coding specific parts of this build was expedient and risky; while writing standard function-scope code has never been easier, the small mistakes it can make conceptually can be substantial to app-wide functionality, and only literate programmers detect and manage them at the warp speed these tools empower.

A specific CSS example: one cannot (safely) place Overflow Hidden on a parent component when programming drag and drop functionality for/across subcomponents. Even if you aren't building out your own drag and drop library, any basic understanding of how dragging and dropping will have to work in a react/css browser environment will govern decisions made elsewhere, and the tools we are using to speed up are not currently keeping these things in mind.

The most valuable a.i. tool I used for this build was voice mode. I went for a run across the Williamsburg Bridge while discussing project scope and dependencies, the folder structure of the build, what new libraries I might consider, any recent updates to keep in mind, etc. When I returned home, my notes were ready for me, along with typescript definitions, and a list of options for state management libraries for me to consider.

Red flags in the race for AGI aside, these are simply... strange times. Technologically accelerated, yet methodologically stagnant. Some devs react to our new dev-speed benchmarks with awe, some with horror. At the very least, the era of aggressive, armchair technologist clients "finding me out" for using an A.i. powered tool seems to be over, with suspicions shifting toward colleagues who refuse to investigate the wisest uses of automation while accounting for associated risks.