In many plants real planning does not happen in the ERP but in Excel — by hand, by work centre instead of by machine, and redone after every change. The ERP gives an order a date without being able to explain why. The course shows how a planner builds a Claude Project that calculates the schedule with code, explains every decision and recalculates in minutes — using rules the plant understands and can change itself.
No programming needed — the rules are written in plain language and Claude turns them into code.
We look at the ERP as a “black box” and at real planning in Excel. Claude Projects basics: instructions, knowledge, conversations; calculations done with code. We collect the minimum dataset — machines, products, routings, calendars, shifts.
Typical errors: missing routings, zero times, duplicates, due dates in the past. Claude proposes, the human confirms — the schedule is calculated only after confirmation.
The plant’s logic becomes written rules: priorities, bottlenecks, buffers, transfer batches (basis: TOC / Drum-Buffer-Rope, Lean). A first weekly schedule by machine; every decision is explainable.
A second shift, an urgent order, a broken machine, more people at a manual station. Answering “when will the new order be ready and which orders does it displace”, comparing scenarios.
Every day the plan is compared with execution: where the data comes from (ERP, manual, QR/mobile), working with incomplete data, a new schedule for the morning meeting, metrics (buffer penetration, late orders).
Who can work on which operation and at what level; the impact of absences, substitution rules, spotting a “single point of failure”.
Serial, one-off, make-to-order, continuous, mixed: how buffers, batches and priorities change; different rules for different departments; planning “learns” from execution data.
Fixed and distance- or batch-dependent transfer time; transfer batch versus full batch. Bill of materials (BOM), synchronising branches to assembly, stocked blanks versus make-to-order.
Minimum batch size, protected orders, customer priority, shift constraints, frozen horizon. A rule is changed in a conversation and applies from the next calculation; traceability of who changed it, when and why.
A realistic roll-out timeline, security and confidentiality of plant data, when a local solution is needed. Defending it to management: hours saved, late orders, transparency. Participants present their projects.