One view of the job
Builders and trades lose margin on variations nobody recorded, not on jobs they underquoted. What to fix first, in what order, and where AI genuinely helps on a building job.
Two complaints usually arrive together. The office cannot say where a job is up to without ringing somebody, and the field has no sensible way to send back what actually happened. They are the same problem seen from opposite ends: the job record is not in one place, so everyone maintains their own version of it.
Variations are where the margin goes
Most builders quote in a spreadsheet built from an older spreadsheet, then retype the accepted quote into accounting. That holds up well enough while the job runs to plan. Variations are where it comes apart: agreed on site, sometimes verbally, captured if at all in a text message or a note in somebody's phone.
At the end of the job, nobody can reconstruct precisely what was agreed and when, so the conversation about the final invoice gets settled at a discount to keep the relationship. That discount is the cost of the missing record.
Most of the margin lost on a job was not underquoted. It was unrecorded.
A variation that is raised, priced, sent and accepted in the same place as the job is the highest-value thing to fix in most trade businesses, and it is well ahead of anything AI-shaped in the queue.
The field needs the shortest path, not the best app
Photos, a note, quantities, a signature, captured on the job and attached to the job. The test is simple: if it takes longer than sending a text message, it will become a text message. It also has to work on a site with no coverage and sync later, because the sites with no coverage are not the exception.
What the office actually needs to see
Not analytics. A list of jobs, the stage each one is at, who owns it, and what is waiting on somebody. Nearly every complaint about having no visibility is answered by that list. Dashboards with charts get built, admired for a fortnight, and then ignored in favour of asking the person who knows.
Subcontractor paperwork is a register problem
Insurances, licences, safety documentation, inductions: collected by chasing people for documents that expire on dates nobody is tracking. A register that knows the expiry dates and chases automatically removes a recurring job nobody wants, and keeps you out of the position of having someone on site whose cover lapsed a month ago.
Where AI fits, and where it does not
It is useful at the edges. Turning site notes or a voice memo into a tidy report somebody will actually read. Reading supplier invoices and matching them to the right job. Drafting a variation description from a few words so it gets written at all rather than remembered later.
It should not price your work, and it should not decide that a variation is approved. Pricing carries your margin and your judgement of the client, the site and the risk, and that judgement is most of what you are selling.
In what order
One job record with stages. Then capture from site. Then quotes and variations in the same place as the job. Then the accounting connection, so the accepted quote stops being retyped. Then, if it still looks worthwhile, the clever parts. Businesses that start with the clever parts usually find themselves back at step one a year later, having paid for both.
The short version
Give every job one record the office and the field both write to. Get variations onto it, because that is where the money leaks. Keep the field path shorter than a text message. Everything else on the list is worth less than those three.
Chasing job status by phone?
Tell us how a job runs from quote to invoice and we will show you where the paperwork is costing you.
Why audit preparation takes a fortnight
For NDIS and care providers, audit pain is almost never about the care delivered. It is about evidence that was never captured in one place, and a process that lives in one person's head.
The gap between finishing a job and invoicing it
In field operations the invoice waits on a job sheet travelling back to the office in a vehicle. Why capturing work on site fixes both the lag and the disputes, and why scheduling is the wrong place to start.