The roadmap is in the deed · owned software · part two
The test was simple. At 4:57 on a Saturday afternoon, two customers try to put a deposit on the same tractor. Who gets it?
Our 593-line plan had never really answered that. It could show the inventory. Send the quote. Take the money. It could also take both deposits — deposits the store’s own paperwork calls non-refundable — and hand somebody a Monday-morning disaster. That is the difference between software that looks finished and software safe enough to run a business.
So I gave the plan to another machine and told it: do not make this prettier. Try to kill it.
It found plenty. A walk-in buys the tractor at the counter on Saturday morning and the website keeps selling it all afternoon. The same internet lead arrives twice and the customer gets two identical opening texts. A provider dies mid-handoff and a “how much is the T474” message never reaches a human at all.
Three assumptions died.
The first plan described what the software should do. The second describes what must happen when it fails.
The first plan tried to build a dealership. The second tried to protect one.
So we tore up the order.
The first thing the system must now prove is one complete loop: inquiry, the right employee, the exact tractor, a correct quote, a deposit, one reservation. One tractor. One winner. One visible record of what happened. Service waits. Parts waits. Financing waits. The big AI promises wait — because a roadmap that cannot say not yet is not a roadmap. It is a brochure.
The new inbox runs beside the old one before it replaces anything. Messages get counted. Inventory gets reconciled. A repeated payment notice cannot create a second reservation. High-risk AI drafts; humans decide. Nothing switches because a date on a roadmap says so. It switches when the evidence says so — and when the owner says go.
That is what ownership actually buys.
With rented software you can request a feature, vote on a feature, and wait for a feature. The roadmap still belongs to the software company. This owner did not buy our first answer — he bought the right to reject our first answer without losing the work already done. He can change the order, stop an automation, refuse a module, bring in another qualified builder, and keep the code, the data, and every good decision already paid for.
Perfect first plans do not exist. The valuable question is what happens when the plan is wrong. Who can change it? Who keeps the work? Who owns what comes next? The first roadmap was ours. The right to tear it up was his.
Code is what you own today. The roadmap is who owns tomorrow.
He owns both.
Bring us the ugly part.
The spreadsheet. The duplicate entry. The quote living in somebody’s texts. The task only one employee knows how to do. We will map the first thing worth owning — and the first way it could fail — before you replace anything.
The 60-second fit check starts it · straight answer