How it works

I come to your shop and do it. Forty minutes if your data is clean, half a day if five years of history is coming across with it.

Your repair order numbers carry on from wherever your old system got to, so the first job written in Teardown follows the last one you wrote anywhere else. Your job stages are named the way your shop names them before anybody signs in.

The watcher goes on the estimating machine at the end of it, and the first estimates are usually across before I leave.

One small program on the machine your estimating software runs on. One download, one run. It checks the export folder every five minutes and sends anything new. The shop does nothing after setup day.

It reads EMS and BMS, from Mitchell and from Audatex, and works out which format a file is by reading it rather than by which folder it came out of.

If you would rather not have a program running on that machine, drag an export folder onto the import screen instead. Same result.

It lands in a queue and counts toward nothing. Somebody moves it to open jobs when the car is actually coming in, and it gets a repair order number at that moment. Nothing goes on the board without a person saying so.

Everything on the estimate comes with it: every line, every part, every labour hour, the customer, the vehicle, the VIN, the insurer, the claim number, the adjuster, the deductible, the date of loss and the estimator who wrote it.

An estimate that is never going to be a job goes to void jobs, where it stays on record, counts toward nothing, and can be brought back.

A revision is matched to the job it belongs to and applied to it. The office gets the difference: which lines were added, which were repriced, which were removed, and what that did to the total.

What your estimating system sent stays exactly as it arrived. A correction made by hand goes onto a new revision rather than over the top of the original, so the figure an adjuster agreed to is still there to point at.

Job stages are named and coloured by the shop. So are the columns on the open jobs list and the order they sit in.

The name and the meaning are kept apart. Renaming Paid to Settled changes the word on the board and leaves the accounting alone.

The spreadsheet you are keeping now comes across as a CSV import.

It shows you exactly what it will create before it creates anything: how many jobs, how many customers, how many vehicles, and every row it could not read. You look at that first, then say go.

Seven CSV files, one for each kind of record, with the row counts beside each one. One click, any time, without asking anybody.

Owner only, because it is the whole shop in seven files.

Shops are separated by row level security in the database rather than by filtering in the application. The rule sits with the data, so a query written wrongly returns nothing instead of somebody else’s cars.

A test suite asserts that one shop cannot see another’s rows. Every negative check is paired with a positive control, so an empty database cannot pass it.

There is no platform administrator back door. Reaching your data needs a time limited grant from you, and every request made under one is logged.

Nightly, to a separate place from the photos.

Each one is verified by reading back what was written, and the restore has been tested rather than assumed.