Why custom software stopped being expensive

A shift scheduling app for a company with a few dozen drivers used to be a project you planned a quarter around. We built one in a week, then rebuilt its reports the following weekend because the manager asked.

What used to make it expensive

Custom software was priced by the hour, and most of the hours went to typing. Every form, every table, every rule about who can see what had to be written by hand, then rewritten when the owner saw it and changed their mind. A tool for a single company could never spread that cost across many customers, so it lost to off the shelf subscriptions almost every time, even when the subscription fit badly.

The typing is now done by AI coding tools, from descriptions we write. The hours that remain are the ones spent deciding, checking and supporting. That is a different, smaller bill.

What takes days now

• Shift scheduling. Drivers see their shifts, request time off, pick up open shifts; managers see each location's board and export a printable week. The rules are specific to one company, which used to be the problem and is now the point.
• Vehicle inspections. A driver scans a sticker on the van, walks around it with the camera, submits. Each submission becomes the next driver's starting point, so damage is tied to the exact shift where it first appeared. Records cannot be edited after the fact, by design.
• Driver handbooks. A login gated, searchable handbook, filtered by location, with each site's own branding, instead of a PDF nobody opens mid shift.
• Permit tracking. A platform that follows a building permit from application through inspections and corrections, for several permit companies at once, each seeing only its own work.
• Lead systems. A CRM shaped around one trade's pipeline, with the website form feeding straight in and alerts to the owner's phone.

None of these was cheap to build a few years ago. All of them now cost less than the meetings about whether to build them.

What still costs money

Be careful with anyone who says software is free now. Three things did not get cheaper:

1. Design. Knowing what the tool should do, and what it should refuse to do. The first version of the inspection app had a screen the drivers hated, and the rebuild took longer than the original. Taste and judgment are still human work, and still the most valuable hours in a project.
2. Data. Getting your existing records in, cleaning them, deciding what is kept and who may see it. A tool built in a day on top of a messy spreadsheet is a messy spreadsheet with a nicer front.
3. Support. Every system needs someone watching it after launch. A supplier changes a login rule, a phone update breaks a permission, a nightly job stops. The cost of fixing is lower now, but the cost of noticing is still a person's attention.

How this changes buy versus build

The old rule was: buy unless you absolutely cannot. Subscriptions were cheaper than anything custom, so you adapted your process to the software. For a landscaper in Palm Beach County that meant paying for a field service product built for a national plumbing chain and using a third of it.

The rule now is closer to: buy the things that are the same for everyone, build the things that are specific to you. Accounting, email, payroll, a payment processor: buy them, they are solved problems with real support teams. The way your crews are scheduled, the way a lead is scored, the way an inspection is documented: those are your process, and your process is now affordable to put into software that fits it exactly.

The question we ask an owner is no longer "can you afford a custom tool" but "is this process worth owning." If the answer is yes, the build is the smaller part of the cost, and the tool is yours to change the moment the process changes.

This is the work we do most weeks now: a scheduling board, an inspection app, a handbook, a permit tracker, each built for exactly one company's way of working, and each kept running afterward by the same team.

Questions people ask