Years ago, I saved principals time by spending a week designing a staff planning tool. Today, AI saves me that week.
In 2019, principals and the HR team at the school district I work for had a hard time keeping up with their staffing plan for the following school year. Some teachers were leaving, some were taking on new roles, some were having a hard time making up their minds, some were looping to a new grade with their students. (In 2020, COVID-19 would hit and make everything a lot weirder, but that's a different story.)
Principals and HR kept each other up to date via spreadsheets, but your typical headaches applied: someone could change a returning staff member in one place but not another; a new version of a spreadsheet forgot to incorporate the most recent changes; updates sometimes happened by email and the spreadsheets could go out of date, etc. Our HRIS couldn't track this well, because these were tentative plans for positions that weren't yet budgeted, not official HR moves.
To help address this, I created an internal website where principals could see their current staff on the left, and their budgeted positions in a grid on the right. They could drag staff members to positions, and visually track who was leaving, who was changing roles, and what vacancies they needed to fill. You can see a sandbox version of it here.
This solved most of the coordination problems above, and folks appreciated the visual cues. This helped us send specific offers to new-to-the-district teachers, and principals could see which positions were open or filled in the spring. (They could also see which of their current staff signed their returning offer letters, and which were holding off.)
We still use the tool, with a few updates, today.
Today I asked Claude Code to create that local-only sandbox. Here's my prompt:
Can you help me build a sandbox of a tool I once built pre-AI? Here's the basic rundown. On the left there's a list of current staff members, with their name/title, big-picture performance (Excelling, Successful, Growing, Struggling) in a colored pill, and which has an independent scroll-box. And then on the right there's a set of positions, governed by big-picture "type", School Leadership, General Education Teachers, Special Education Teachers, Student Support, Administrative Support, etc. And teachers can be drag-and-dropped into one of those positions or into a "not returning" bucket at the end. Except for not-returning, all are one-person/one-position matches. And dragging a new person into a position puts the old person back on the left menu.
What this solves for schools is having a good sense of what their next school year will look like, so they can figure out what specific vacancies they'll have, and plan for growth/etc.
Make sense? Ask me questions if needed. I believe this used dnd-kit originally, but no need to stick to the legacy version since this isn't a true reproduction.
It had no questions for me. A couple minutes later, it produced the demo I linked above.
To be fair, the real version of this website has a number of other things going on: it pulls a live staff list in from our HRIS, sends and tracks offers through an e-signature tool, and has a few layers of authorization and permission checks. But, reader, it took time for an experienced professional (me) to get the drag-and-drop working just right.
I also never figured out how to support it on a mobile screen. Phones are too small to display the staff list and the grid. The real app just tells our principals to open the page on a laptop. When I presented the issue to Claude, it suggested and implemented a tap-to-assign feature that makes the phone experience totally functional. Again, Claude needed little from me beyond waiting for a minute or two.
Where does this leave (former) expertise?
I'm impressed by what AI can now easily do. What took a specialist days can now be implemented by any English speaker and an AI assistant in minutes. But I'd be lying if I denied that I'm a bit disheartened, too. I spent a lot of time learning to read and write the appropriate languages and libraries, and I've been proud of coming up with, and implementing, clever solutions to people's problems.
Still, I think there's an art and expertise in layering prompt after prompt into a sustainable architecture that actually saves time and can adapt to the next set of questions. For now, anyway.
And AI collapsing the need for expertise may also open new doors. School systems, as a rule, don't hire software engineers. They're expensive and leave organizations in a bind when they resign and take their knowledge with them. So we buy off-the-shelf software, which is often pretty janky, but which comes with an account executive and a support email.
But if the pool of app creators expands from trained programmers to lots of technically minded professionals with an AI subscription, maybe school systems and similar organizations can look to a future where they have more software that fits their problems, because they prompted it themselves.