Maintenance-Due Forecaster: Service on the Real Interval, Not Too Early or Too Late
Combine calendar intervals with usage meters (run hours, cycles, mileage) to project when each asset is truly due — whichever comes first — then let a planner approve the due list before it goes to scheduling.
A web tool where you import assets (with both time intervals and usage thresholds) plus meter readings, and it projects each asset's next due date — calendar-due or usage-due, whichever comes first — using an estimated usage rate from reading history. It flags assets with stale readings, a planner reviews and approves the forecasted-due list, and approved items hand off to scheduling. Everything exports to CSV.
Before you start
- A Supabase account (free)
- A Vercel account (free)
- A Resend account (free)
- A CSV or Google Sheet of your assets with calendar intervals and usage thresholds
- A CSV or Google Sheet of your latest meter readings
- Claude Code or any AI coding agent
The problem this kills
Preventive maintenance lives in two worlds and they never line up. Some service is on the calendar — every 6 months, every quarter. Some is on usage — every 250 run hours, every 10,000 cycles, every 5,000 miles. So you either service everything on the calendar (and over-maintain the machines that barely run while ignoring the one that's redlining), or you chase meter readings in a spreadsheet and miss the calendar items entirely. Either way you're guessing, and "guessing" on a critical asset means a breakdown you'll pay for at 2 a.m.
The honest answer for any asset is: it's due on whichever trigger fires first — the calendar date or the projected date it crosses its usage threshold. Working that out by hand for hundreds of assets, with meter readings that arrive at irregular times and sometimes go stale, is exactly the kind of tedious math a small internal tool should do for you. You don't need to be a developer to build it.
What you'll build
A simple internal web tool for your maintenance-planning team. You import your assets — each with its calendar interval (e.g. every 180 days) and its usage threshold (e.g. every 250 run hours) — and you import your latest meter readings from a CSV or Google Sheet. The tool estimates each asset's usage rate from its reading history (hours per day, cycles per week, miles per month), projects the usage-due date, compares it to the calendar-due date, and reports the earlier of the two as the real due date. It flags assets whose readings are stale (so it isn't forecasting off old data) and warns you when a projection is low-confidence. A planner reviews the forecasted-due list and approves which assets to schedule — the human gate — and approved items hand off to scheduling and export to CSV in the exact columns your scheduler or CMMS expects.
What's inside the Implementation Plan
The downloadable plan is a step-by-step file you paste into an AI coding agent. It opens by interviewing you about your business — what kinds of usage you track (run hours, cycles, mileage, something else), how your intervals and thresholds are written today, exactly what your asset and meter-reading exports look like, how often readings come in and from where, your real asset IDs and naming, what counts as a "stale" reading, who is allowed to approve the due list, and the messy edge cases like brand-new assets with no reading history or meters that roll over or get reset. It reads a short spec back to you for a thumbs-up, then builds the forecaster around your assets and rules instead of a generic template. From there it walks the agent through the data model, the asset and reading imports, the usage-rate estimator, the whichever-first due-date engine, the stale-reading flags, the planner's approval gate, the scheduling handoff, and the CSV export. Every step ends with a ready-to-copy prompt.
The governance it includes (this is the point)
This isn't a toy. The plan builds in the controls a real maintenance function needs: login so only your team can use it, row-level security so people only see their own organization's assets and readings, a complete audit trail of every import, forecast run, approval, and handoff (who, what, when), a hard human-approval gate so no asset goes to scheduling until a planner signs off on the forecast, and duplicate guards — a forecast is deduped on (asset + maintenance cycle) so the same service can't be queued twice, and re-importing the same reading file is blocked. The tool exists to make a careful human decision easy: the math proposes the due date, a planner decides what actually gets scheduled.
Who it's for
Maintenance planners, reliability leads, and asset controllers who are tired of either over-servicing machines on a blanket calendar or getting caught out by the one asset that runs twice as hard as the rest. If you can describe how you set an interval and what your meters measure, you can build this.
You've got this — open the plan, paste the first prompt, and you'll have a real due-date forecast for your own assets by the end of the weekend.