An equipment ledger for an event decorator
Tracking what's gone out to which wedding — mandap frames, LED strings, sound systems — where the shared spreadsheet has started fighting back.
Several people edit one spreadsheet of loaned-out equipment and it has begun to cost more than it saves.
Bring the spreadsheet with you
The single most useful thing you can do here is describe the columns you already have. You have already done the data modelling; you just did it in a spreadsheet.
We are Utsav Decor, event decorators in Nagpur. We track equipment loaned out per event in a spreadsheet with these columns: item, code number, which event it's gone to, who took it, date out, date due back, condition notes. Six people edit it and we keep overwriting each other. Build the same thing as a web app where each person signs in, and show me what's due back and what's overdue.
"Six people edit it and we keep overwriting each other" is the most important sentence in that prompt. It tells the agent this is a multi-user problem, which it will otherwise not assume.
What it did
Sign-in, an equipment table, a loans table linking each item to an event and a person with dates, a dashboard defaulting to what's overdue, and a per-person view. It correctly modelled a loan as its own record rather than as columns on the equipment row — which is what lets you see, months later, every wedding a mandap frame has been to.
It also put the condition note on the loan rather than the item — so you can see what state something came back in each time: which LED strings came back tangled, which frame came back bent. Nobody asked for that.
Then import what you already have
Add a way to paste in rows from our spreadsheet to create equipment records in bulk. If a code number already exists, update that record rather than making a duplicate.
Migration is the step people forget, and it is the step that decides whether the tool actually replaces the spreadsheet or just sits next to it.
What we had to correct
Every guide on this site has this section. If it were empty, the guide would be a demo.
Anyone could sign in. It built authentication — a real sign-up form, properly hashed passwords — and left it open to the internet. Six people at Utsav Decor were meant to use this. We had to say so explicitly: "only these email addresses should be able to create an account."
This is the same class of mistake as the public signups list in the pickle-brand guide, and it is the one to watch for. The agent builds the mechanism you asked for and does not infer the boundary you didn't mention.
Deleting equipment deleted its loan history. Technically correct — the loans referenced the item — and completely wrong for the job: retire one bent mandap frame and every event it had ever been to went with it. We asked for equipment to be archivable instead, which is what we should have said the first time.
Items due back today were marked overdue by breakfast. The app treated the due date as ending at midnight on the server's clock, which runs five and a half hours behind India — so anything due back on the 20th went "overdue" at half past five that same morning, and a crew returning sound systems at nine was told they were late. One request fixed it — work out due dates to the end of the day, Indian Standard Time — but note why it happened: nothing in our prompt said where we were, and a date with no place attached belongs to the server's clock, not yours.
Where it stopped short
No audit trail — you can see the current state, not who changed what when, which for a tool about who-has-what is a real gap we'd close next. No reminder goes to anyone before a return falls due. And ticking items back in still means standing at the godown with a laptop, when the person actually doing it is on the back of a truck with a phone.
It replaced the spreadsheet, which was the bar. It didn't become software you'd sell, which was never the bar.