Every job that runs through a shop leaves a trail. What it was quoted at. What it actually cost to run. Which customers pay on time and which ones chase you for it. Which jobs blow past estimate and why. If you run JobBOSS2 — or almost any job-shop ERP — that trail already exists. It's sitting in your system right now.
The problem is nobody's asking it anything.
Why "just ask Copilot" doesn't work
Microsoft 365 Copilot is good at reading what it can see — email, Word docs, Excel files, SharePoint, Teams conversations. It has no idea your ERP exists. JobBOSS2 isn't a Microsoft product, it doesn't sync to your mailbox, and Copilot has no native line of sight into your quoting history, your job costing, or your floor data.
So when someone tries "Copilot, which customers are least profitable this year" or "summarize on-time delivery by month," it comes back empty. Not because the answer doesn't exist — because the question never reaches the data.
That gap is where most shops give up and go back to pulling reports by hand.
What's actually sitting there, unused
Think about how much of this only gets looked at once, if ever:
- Quote-to-actual variance, job after job — the single best predictor of where estimating is off
- Customer-level margin, not just revenue — who's actually worth the capacity you're giving them
- On-time delivery trends by customer, by part, by season
- Rework and scrap patterns that never make it past the shop floor
Most of it lives and dies inside the ERP. It gets pulled into a report maybe once a quarter, if someone has the time. The rest of the year, it just sits there.
The data was never the problem. Getting it somewhere Copilot can actually read it — that's the whole job.
Where shops get this wrong
The instinct is to treat this as a settings problem — find the right toggle, connect the app, done. It isn't. JobBOSS2 doesn't hand its data to Copilot out of the box, and the shops that try to force a shortcut usually end up with a reporting mess that's harder to trust than the manual process they started with: numbers that don't reconcile, a "dashboard" nobody looks at twice, another subscription that quietly stops getting used.
The shops that get real value out of this do the unglamorous part first — deciding what data actually matters, getting it into a form a tool like Copilot can reliably use, and keeping it in sync without someone babysitting it by hand. That's a real project. It's not a weekend, and it's not a plugin.
This is exactly the kind of problem Rahn works on with manufacturing clients — bridging a shop system like JobBOSS2 with the AI and reporting tools you already pay for, so the data you're already generating actually gets used. Not a rebuild, not a new platform to learn. A way to ask the questions you're already asking by hand, and get an answer in seconds instead of an afternoon. If this sounds like your shop, the Technology Assessment is the place to start — one session, a clear picture of what's connectable and what isn't, and a straight answer on whether it's worth doing.
The takeaway
The data your shop needs is already there. The gap isn't information — it's the bridge between where that data lives and where you'd actually use it. Close that gap, and Copilot stops being a novelty and starts being a second set of eyes on the numbers you're already generating.