Walk through any corrugated plant‘s front office this fall and you’ll find someone pasting a spreadsheet export into a general-purpose AI tool. They might be pricing a reorder, hunting for margin leaks, or trying to explain last week’s waste on the flexo folder-gluer.
That isn’t a prediction. It’s what operators said on stage this month. At the Amtech User Conference panel, covered by Board Converting News, one industry chairman described team members running ad hoc analyses “with the LLM of your choice,” analyses that are surfacing pricing, margin, labor and waste-reduction opportunities that used to require a seasoned analyst. A CTO from a 30-plant group described employees building “the perfect dashboard” for themselves, and then the hard part: IT having to make it supported, governed and scalable.
Call it shadow AI. The productivity is real. So is the risk.
The problem isn't AI. It's ungoverned AI.
Every copy-and-paste into a public AI tool is plant data leaving your control: customer pricing, board costs, order history. And it’s happening in the industry attackers target most. Manufacturing remains the most-hit sector for ransomware this month, and packaging companies keep turning up on leak sites.
There’s also an accuracy problem. A spreadsheet exported on Monday is stale by Wednesday. AI working on a stale copy will give you confident answers that are wrong. As one panelist put it, AI output “will look incredibly convincing,” so an expert still has to sign off.
Where AI belongs: inside the system of record
The fix isn’t banning AI. It’s bringing AI to your ERP, under your rules, instead of carrying your ERP data out to AI.
That’s the direction Amtech is building in. Amtech Intelligence puts AI inside the EnCore ERP and Amtech MES screens your team already uses. In Amtech Quick Quote, an AI agent already reads incoming RSC and HSC quote requests in whatever format customers send them. It drafts the spec, flags anything missing, shows how confident it is in each field, and waits for an estimator to approve it.
Next, Amtech is delivering MCP (Model Context Protocol) to connect EnCore to the AI assistants your people already use. It runs through governed access with your permissions, not uncontrolled exports. Every Amtech agent follows the same rule: agents propose, people approve, and your data stays under your control. With hosting on AWS and a security program built on the NIST and ISO 27001 frameworks, IT gets the oversight it needs without slowing the business down.
Compare that with the alternatives being marketed right now: chatbots bolted on beside the ERP, or “AI-ready” platforms that require a full re-implementation first. Forty years of building software for box plants has taught us that new technology sticks when it fits the workflow people already run.
The fix isn’t banning AI. It’s bringing AI to your ERP, under your rules, instead of carrying your ERP data out to AI.
That’s the direction Amtech is building in. Amtech Intelligence puts AI inside the EnCore ERP and Amtech MES screens your team already uses. In Amtech Quick Quote, an AI agent already reads incoming RSC and HSC quote requests in whatever format customers send them. It drafts the spec, flags anything missing, shows how confident it is in each field, and waits for an estimator to approve it.
Next, Amtech is delivering MCP (Model Context Protocol) to connect EnCore to the AI assistants your people already use. It runs through governed access with your permissions, not uncontrolled exports. Every Amtech agent follows the same rule: agents propose, people approve, and your data stays under your control. With hosting on AWS and a security program built on the NIST and ISO 27001 frameworks, IT gets the oversight it needs without slowing the business down.
Compare that with the alternatives being marketed right now: chatbots bolted on beside the ERP, or “AI-ready” platforms that require a full re-implementation first. Forty years of building software for box plants has taught us that new technology sticks when it fits the workflow people already run.
Why it matters
- Owners/CEOs: Margin insights stop depending on one analyst’s spreadsheet.
- CIOs and IT leaders: You move from blocking AI to governing it.
- Plant managers: Answers come from live data, not last week’s export.
What to do today
- Run a 15-minute shadow-AI audit. Ask your estimators, CSRs and schedulers which AI tools they used this month and what data they pasted in. Don’t punish anyone. Map it.
- Pick the one use case worth governing first. For most box plants, that’s the quote request inbox. Count how many emails it took to produce yesterday’s quotes and how long each one waited.
Ready to bring AI inside your plant data, safely?
Talk to an Amtech expert about Amtech Intelligence, AI-powered quoting in Amtech Quick Quote, and the EnCore MCP roadmap.
