Custom software for manufacturers, without the spreadsheet sprawl

It is 6am and the line lead is reading production numbers off a whiteboard. The office is reconciling those numbers against a spreadsheet someone emails around at noon. The ERP knows a version of the truth, the MES knows another, and the real state of the floor lives in three people's heads and a stack of clipboards. Everyone is busy, nothing is wrong on paper, and yet nobody can answer a simple question quickly: what did we actually make yesterday, and where is it now.

That gap is where custom software earns its keep. Not a flashy rip-and-replace, and not another off-the-shelf system you bend your whole operation around. Just software that fits how your plant already runs, pulls the scattered numbers into one place, and stops asking people to retype data that a machine already knows. That is the shape of our custom software work for manufacturers.

Start with the data you already have

Before anyone writes a line of new code, a good team goes looking for where the truth currently lives. In a manufacturer that is almost always more places than you would like: an ERP, maybe an MES, a few PLCs and SCADA exports, a barcode scanner system, and the inevitable spreadsheets that quietly became load-bearing.

The first job is not to replace those systems. It is to read them honestly. What does the ERP consider a finished good? Where does inventory get decremented, and does that ever match the physical count? Which "report" is really just one person's macro-laden workbook? The database and the exports are the truest record of what your plant actually does, and they tell a clearer story than any vendor brochure.

Out of that you get a plain map: what each system owns, where the numbers disagree, and which manual steps exist only to paper over a missing connection. That map is the spec. Skip it, and you build a shiny dashboard on top of numbers nobody trusts.

Integrations are the whole game

For manufacturers, custom software lives or dies on integration. The value is rarely a brand-new system standing alone; it is the layer that finally gets your ERP, MES, and shop-floor devices speaking the same language so a part scanned at a station updates inventory, work orders, and the office dashboard at once, without a single retyped number.

That is careful API and backend engineering: reading from systems that were never designed to be read from, normalising units and part numbers that drift between databases, and writing back without corrupting the system of record. Some vendors expose a clean API; many expose a flat-file drop, an ODBC connection, or a port on an aging server. A good team meets each one where it is and wraps it in something dependable.

The connective tissue between all of it is plain automation and integration work: the scheduled syncs, the event handlers, the reconciliation jobs that catch the day the MES and ERP disagree and flag it instead of silently picking a number. This is the same discipline that keeps complex logistics operations honest, where a wrong count downstream becomes a truck in the wrong place. Done well, it is invisible. Nobody notices the integration that just works, which is rather the point.

Dashboards for two very different audiences

The shop floor and the office want opposite things from the same data, and pretending otherwise is how dashboards end up ignored.

On the floor, the screen needs to be glanceable from across a bay, honest about right now, and forgiving of gloves and grime. Big numbers, clear status, current run rate against target, the next changeover, anything blocking the line. No menus to dig through, no login dance every shift. If an operator has to interpret it, it is too clever.

In the office, the same numbers feed planning and accountability: throughput by line and shift, scrap and rework trends, OEE that someone actually trusts, schedule attainment, inventory that ties out to what finance believes. Here the work is correctness and history, not glanceability. A real-time number on the floor and a defensible monthly figure in the office come from one pipeline, presented two ways. Built on modern Next.js patterns, the floor view stays live and the office view stays exportable, from the same source of truth.

Replacing the aging on-prem system, carefully

Most manufacturers are not starting from nothing. They are running a system from a vendor who got acquired, on a server in a closet, that only one retiring person fully understands. The instinct is either to nurse it forever or to rip it out in one terrifying weekend. Both are bad.

The calmer path is a migration and replatforming approach that runs the new alongside the old. You start by reading from the legacy system and surfacing its data in the new dashboards, so the old box keeps being the source of truth while you build confidence. Then you move one workflow at a time, reconcile constantly, and only retire the old system once the new one has quietly matched it for weeks. No big-bang cutover during a production week. No month where nobody knows which number is real.

That patience matters more here than almost anywhere, because the cost of a wrong number is not an awkward report. It is a missed shipment, a stockout, or a line idled for parts.

Reliability is a feature, not an afterthought

A consumer app can have a rough night. A plant cannot. If your production tracker goes down mid-shift and people fall back to clipboards, you have not just lost a tool, you have lost the trust that made them stop using clipboards in the first place.

So reliability gets designed in from the start: the floor view degrades gracefully when a network blips, syncs queue and catch up instead of dropping data, and the system fails loud rather than silently showing yesterday's numbers as if they were today's. None of this is glamorous. It is the boring, unbreakable plumbing that lets people stop double-checking the software and get back to running the plant. That is the good kind of lazy.

The takeaway

Custom software for a manufacturer is rarely about one heroic new system. It is about reading the data you already have, connecting the systems that refuse to talk, and presenting one honest set of numbers to two audiences who need them differently. Do that on top of careful integrations and a patient migration, with reliability treated as a real requirement, and the spreadsheets quietly retire themselves.

This is the work we like: the unglamorous plumbing that turns three versions of the truth into one. If you are staring at an aging system, a wall of spreadsheets, or numbers that never quite tie out, tell us what your plant runs on and we will give you a straight read on what it would take to make it boring and reliable.