Your Data Was Never the Problem

August 24, 2026

You’ve been told for a decade that your data is why you can’t do any of this, that the analytics and the automation all wait on a cleanup you’ll never finish. It’s a good excuse, and it just expired.

Start with the part nobody says out loud: Your ERP data is fine. You run the whole plant on it, every order, spec, schedule, and invoice, and you bet the business on it every day. The mess was always the stuff around it, the shared drives, the spreadsheets, the one-off exports. And what stopped you was never that mess. It was the price of using it.

Using it meant a project. Stand up a data warehouse, hire someone to haul your numbers into their platform, pay them to clean it there, build the dashboards, and rent you a chatbot to ask questions of your own data. Six figures and a year, and you owned none of it at the end. We all know how that goes: the kickoff, the discovery phase, the scope creep, the go-live that slips two quarters, and a dashboard nobody opens because the numbers are already stale. That on-ramp is what kept you parked, and “we’ll start our data initiative next year” has been the answer for 10 years running.

The On-Ramp Just Collapsed

It collapsed from two directions at once. Your ERP vendor is shipping AI that reads its own copy of your business
data. Amtech folded Clarico, a machine-
learning estimating and quoting engine, into the system, and Advantive switched on AIVA across its product line, Kiwiplan included, so you can put a plain-English question to your operational data instead of waiting on a report. No project. The quoting and costing brains stay where they belong, inside the system that already knows your board combos and your corrugator schedule.

Microsoft is shipping the other half, the fix for the messy data around the edges. The stack most plants already pay for and underuse will pool your scattered files into Fabric, let Copilot shape and check them, answer questions in the same Teams window your team already lives in, and build the Power BI reports. Fabric’s data agents do exactly that, in preview today. That bundle, the warehouse, the cleanup, the dashboards, the chat, is the entire business those consultants used to sell you, now running on software you own.

Connect the two halves, your ERP’s AI on the operational core and Microsoft’s on everything else, and you land where the yearlong warehouse project was supposed to put you, minus the project. What used to cost a year and a consulting retainer now costs a capacity license and a few weeks of someone’s attention. That’s the genuinely new part. You can get there from where you already sit, with the people and budget you already have.

Don’t read that as Microsoft doing all of it. It will even sell you a quote agent: Its Business Central Sales Order Agent reads requests out of a shared mailbox and drafts the order today, but only for shops on its generic ERP. A box plant gets nothing from it, because Business Central can’t quote a box. The office half is Microsoft’s. The quoting half stays in your system, and it takes both to finish the job.

Quoting Back on the Hard Jobs

Every piece of an automatic quote already sits in your plant: a portal to take the request, an engine to price it, a scheduler, a trigger that orders paper and tooling. Amtech’s customer portal already prices common reorders off a built-in matrix and books them in one click, no CSR, no email. The front door works. What it couldn’t do was the oddball job, the one that needs a designer and an estimator, because wiring that last stretch used to mean the project. Now it doesn’t. Your customer gets a price while they’re still on your site, and the job hits the schedule before your competitor has called them back.

It still isn’t push-button. Nobody’s AI cleans your data while you get coffee, and somebody on your side has to learn the tools and decide what “correct” looks like before a machine can hold to it. But closing that gap is now an afternoon with a tool you own. It used to be a year and a vendor you rented. The work moved from “fund a project” to “pick it up,” and that’s the whole difference.

The shops that move now will be the ones that quit waiting for the data to clean itself or a consultant to come fix it, and start learning the tools already sitting in front of them. The rest will keep saying “next year,” the way they have for 10, while the plant down the road wires its front door to its back office and starts quoting in minutes.