The hard part is the data underneath.

Standing up an acquired business as a new company inside an existing ERP: org structure, general ledger, AR and AP, sales classes, units of measure, item and inventory master, warehousing, work centers, routings, and security. Including the decisions that are expensive to reverse later — shared versus separate master tables, item numbering, and the cutover date.

ERP data is only useful when someone can see it. I build the reporting layer on top — operational reports out of the system itself, and dashboards in Qlik or Power BI that pull from live ERP data rather than from a spreadsheet somebody maintains by hand. Margin by product line, cost per unit across plants, machine utilization, on-time delivery.

I look at what a company already has — the ERP, the data inside it, and the people who know how the business actually runs — and find where AI can carry real weight. Usually that is not a chatbot. It is a layer on top of the ERP that puts the right numbers in front of the right person at the moment they are deciding something.

The point is not to replace the people who know the business. It is to put what they know together with what the system knows, so a call that used to take a week of pulling reports takes an afternoon, and it is made on numbers rather than instinct.

That only works if the data underneath holds up, so I check that first. I also cover the governance question most assessments skip — which data is allowed to leave the building — and I use AI in the build itself, which takes real hours out of a project.

Getting real numbers off the plant floor and into the system: data collection across presses and finishing equipment, labor and material capture, budgeted hourly rates validated with the people who run the machines, and unit costs that hold up when someone checks them against actuals.

Deep, hands-on experience with the print and packaging stack — estimating front ends, scheduling, and shop floor collection bolted onto a commercial print MIS to make it work for packaging. I have architected that configuration from the ground up, including the seams between products that no single vendor owns.

The same work outside of print. Discrete and light manufacturing where product moves through more than one system — item and BOM structure, inventory across locations, order-to-cash, and the integrations between the ERP and whatever else the business runs on. Often that means a manufacturer selling through wholesale, retail, and direct-to-consumer channels at once, where cost and inventory visibility has to reconcile across all of them.

The layer that decides what to buy and when. Material requirements planning driven by real lead times, safety stock, and reorder points rather than a buyer’s memory. Demand forecasting models that account for seasonality and history, so inventory is sized to what is actually coming. And warehouse management underneath it — location and bin structure, receiving, pick and pack, cycle counting, and the handoffs between the warehouse and the ERP that usually get skipped.

The connective work around an ERP: site-to-site networking for multi-plant data, reporting rebuilt against a new schema when a legacy system goes away, and automation for the manual steps that survive a go-live because nobody had time to fix them.