Build your own AI Performance Manager in under 10 minutes
Install an experienced fixed-ops manager into Claude or ChatGPT. It reads your invoice export, audits it before it trusts it, and hands you the analysis a good multi-store operator would give you after a weekend in your numbers.
Download the free starter kitFour steps. You do this once.
After setup it's one file drop a month — and it gets more useful over time, because your project remembers.
Export your invoice detail
The kit shows you what to pull from your shop management system, and how far back to go.
Create a Project
One click in Claude or ChatGPT. A project is just a workspace that remembers.
Paste and upload
Copy the instructions in. Add the four knowledge files. Nothing to edit or fill in.
Attach your export and ask
Type “Analyze my shop.” That's the whole thing.
What's in the kit
Nine files. The instructions are the engine; the rest is what your AI knows once it's running.
Your shop speaks its own language
Your AI will hit exactly one wall, and it's worth understanding before you start, because it's the one thing a clever prompt can't get around.
Nobody types service names the same way twice. Your advisors have their own shorthand, your vendors push their own branded names, and your system has been collecting all of it for years. So when your AI goes looking for oil changes, it finds this:
Count those naively and a shop that sold 40 oil changes reports well over a
hundred. It works the other way too: a shop writing
BG BRAKE SYSTEM FLUSH — FLUSH ENTIRE BRAKE SYSTEM on 109 repair
orders is invisible to anything that only knows BRAKE FLUID FLUSH.
Same service, different words, 109 units uncounted.
The kit handles this as honestly as anything can. Your AI is told never to categorise by keyword, to read the descriptions properly, and to hand you a section called “Your Shop's Dialect” — the phrases unique to your shop, what each one is worth, and how confident it is that it read them right.
And this costs more than a tidy report. A vehicle's service history is what tells you how to advise its owner. If that brake fluid service was written in a way nothing recognises, the vehicle reads as overdue for work it already had — so you recommend it again, in front of a customer who remembers having it done, or you miss the service it genuinely needs. Both cost trust at the counter, and trust takes longer to rebuild than revenue.
One thing worth saying plainly: this is not something to go and fix. It isn't a discipline problem and it can't be solved at the keyboard. A real history runs to hundreds of distinct descriptions, new ones appear every time a vendor renames a product or you hire an advisor, and nothing lets you retype the last ten years. A mapping spreadsheet is stale the week after you build it, and then you have two versions of the truth instead of one. Don't send your team on that errand.
Measure it, know the number, and treat it as what it is — the honest edge of what your raw export can prove on its own.
Get the starter kit
Free, and yours to share with anyone who runs a shop.
What this can't do
Worth knowing before you start, so the first report doesn't surprise you.
It only knows what you sold
An invoice export can't tell the difference between work a customer declined and work nobody offered. Both look identical — no sale. One is a closing problem, the other is a process problem, and they need opposite fixes.
Your service descriptions are free text
Your system probably writes the same job a dozen different ways. Your AI will group them as best it can and tell you how confident it is — but every category number rests on that guesswork. What it says about this is usually the most revealing thing in the report.
Nothing happens on its own
You have to remember to pull the export, start the chat and ask. There are no alerts, nothing lands in your inbox at 6am, and nobody gets told when a good customer goes quiet. It's a very sharp tool that only works when you pick it up.
Then the obvious question shows up
Somewhere in your first report, your AI will tell you how many of your service descriptions it had to interpret, and how much revenue sits behind the ones it wasn't sure about.
That number is the ceiling on everything else. Standardizing service history across every shop management system is the genuinely hard problem in this industry, and it's the one SideKick360 exists to solve — so a vehicle's history is actually known, and what it's due for can actually be trusted.
That's the difference between a report you read and a conversation your advisor can have at the counter, holding a vehicle's real history, recommending the service it actually needs.