Planned-Maintenance Notice Scheduler

The problem this kills

Scheduled downtime is supposed to be the easy kind of downtime - you planned it. Yet it still turns into angry tickets, because the notice went out too late, went to the wrong people, or told a customer in Sydney that maintenance starts at "2 PM" without saying whose 2 PM.

Most teams run this out of a shared inbox and a calendar reminder. Someone hand-writes the heads-up, BCCs a list they hope is current, and tries to remember to send the "starting in one hour" follow-up. When the window moves - and it always moves - the reminders that were already queued go out anyway, and now you've told everyone the wrong time.

This tool turns that scramble into a repeatable, governed process: one maintenance event, the right segments, a reminder series that fires automatically at the lead times you choose, each one written in the customer's own local time. Reschedule once and every pending reminder updates itself.

What you'll build

An internal web app where your support and ops team can:

• Define a maintenance window (start, end, scope, affected services) once.
• Select the customer segments it actually affects from your imported list.
• Draft the notice once - the tool fills in each recipient's local start and end time automatically.
• Send it to a teammate for a human approval before anything is scheduled.
• Have Resend send a timed reminder series (e.g. 7 days / 1 day / 1 hour before) in each customer's time zone.
• Reschedule or cancel in one place, which updates or kills every pending reminder.
• See a complete send log and queue an automatic completion / all-clear notice when the window ends.

What's inside the Implementation Plan

A single markdown file you paste into Claude Code. It walks the AI agent through building the whole tool, step by step, with a copy-paste prompt at the end of each step.

Crucially, the plan opens by interviewing you about your business - your current maintenance-comms process, the systems and spreadsheets you use, how your customer/segment data is actually shaped and named, your typical and peak volumes, and your real approval rules and edge cases. It then reflects a short tailored spec back to you for a thumbs-up, and shapes the data model, the reminder logic, and every later step around your answers. You get a tool fit to how your team actually works, not a generic template.

You don't need to know how to code. You answer questions in plain language and paste prompts.

The governance it includes (this is the point)

This is built like an internal system of record, not a toy:

• Login so only your team can use it.
• Row-level security so each organization only ever sees its own maintenance events and customer lists.
• A full audit trail - who created, edited, approved, rescheduled, or cancelled each notice, and when.
• A hard human approval gate - the AI drafts the notice and the recipient list, a person reviews and approves, and only then is the reminder series scheduled. Nothing is sent on a draft.
• Duplicate guards - a dedupe key of customer + maintenance event + reminder stage means the same person can't get the same reminder twice, even if you reschedule.

Who it's for

Ops, support, and reliability teams who coordinate scheduled downtime - SaaS support leads, IT service desks, managed-service providers, and anyone who has to tell customers "we'll be down for an hour" without it turning into a fire drill.

You've got this - paste the first prompt and let the agent interview you.