We have been shrinking for forty years. Is spending on software sensible?
Only if it pays back inside a year, and that is the test we would hold it to with you. In shops this size the payback usually comes from two places: quotes going out before the customer gives up waiting, and finding out which job type has been losing money quietly. If we cannot see a credible path to either, we will tell you not to spend.
Everything runs on paper travellers. Does that all have to change?
No. Paper on a shop floor works, it survives coolant, and your people are used to it. What we usually do is keep the paper and photograph it, so the record ends up digital without changing the habit. Forcing a machinist onto a tablet mid-shift is how these projects fail, and we would rather work with what you already do.
Our best mechanic is retiring next year and it is all in his head. Can anything be done about that?
Some of it, if you start now. What can be captured is the machine-by-machine record — what was fitted, what failed, what the workaround was. What cannot be captured is his judgement. Being honest about that split is the difference between a useful project and an expensive disappointment, and we would rather set the expectation properly.
How is a build paid for when our money follows a purchase order?
Fixed price against a written scope, with the stages tied to delivery rather than to a calendar month. If your cash arrives when a customer pays a PO, we can align the schedule to that. What we will not do is quote a range and let it drift, because in a shop this size a drifting number is the whole margin.
If we ever stopped, do we keep the job history and the cost data?
All of it. Quotes, actual hours, inspection records, machine histories, exported in ordinary formats. It is written into the agreement before you commit. In a shop that lives on knowing what a part really costs to make, that history is worth more than anything we would build to hold it.
Is this really AI, and does that matter for what we do?
The AI part is reading a drawing or a handwritten traveller and turning it into data. That is a genuinely hard problem and it is where the hours come back. Everything after that — comparing to past jobs, ageing an unquoted enquiry, flagging an inspection — is ordinary programming and we charge it as ordinary programming.
Who checks a quote before it reaches a customer?
You do, and only you. Our review gate is SolaceSentry and it will not release a quote, an invoice or a certificate until a named person approves it. Given that a price sent by mistake in this trade can cost more than the job earns, that gate is not negotiable and we would not build without it.
We work jobs in both Henry and Madison counties. Does that complicate things?
It complicates your week more than it complicates the software. Being on US 36 means half your work is over the line, and it is the fastest way to lose a crew day. Once a job knows its own address, it can carry the right county's requirements from the start, and the two sets of inspection paperwork stop landing in the same pile. That single distinction is worth more than most of what people ask us for.