Production Planning delivers the complete MRP II planning stack as a single, closed loop. It ingests and shapes a demand picture through statistical forecasting, blending historical consumption, seasonality and trend with committed sales orders and manual overrides into a consensus demand plan. From that demand it maintains a master production schedule, the rate-and-timing statement of finished goods that anchors everything below it. Material requirements planning then explodes that schedule against product structures to derive time-phased net requirements for every component and raw material, netting against on-hand stock, open purchase orders and open work orders, and generating planned orders with correct lead-time offsets.

Capacity planning runs in the same pass. Rough-cut capacity planning validates the master schedule against key resources before commitment, and detailed capacity requirements planning loads every planned and released order onto work centres to expose overloads. Finite scheduling then sequences work against genuine, bounded capacity, respecting shift calendars, resource availability and constraints so the resulting plan is executable rather than merely arithmetically balanced. The module maintains this plan continuously, regenerating on change and surfacing the exceptions that need a planner's judgement.

At the heart of the module is the idea of demand and its progressive refinement into a commitment. Raw demand arrives as forecast and as firm orders; the planning process reconciles the two so that a forecast is consumed by actual orders as they land, avoiding the double counting that inflates plans. This reconciled demand feeds a planning horizon divided into time buckets, and the entire model is fundamentally time-phased: quantities are meaningful only in relation to the period in which they are needed or available.

The second central concept is the plan itself, expressed as a hierarchy of scheduled quantities. A finished item carries a rate and timing at the top; beneath it, the product structure describes what it is composed of and in what proportion, and the routing describes the sequence of operations and the resources each consumes. Planning walks this structure to translate a demand for the top-level item into dependent demand for everything below it, offset by the lead time each level requires.

The third concept is capacity, modelled as the finite, calendar-bound availability of the work centres and resources that perform the operations. Planned quantities become a load against that availability, and the tension between required load and available capacity is what the module exists to resolve. Supporting these are planning policies, the rules attached to each item that govern how it is replenished: lot sizing, safety stock, lead times and the sourcing decision that says whether a shortfall is met by making or by buying. Together these concepts let the module reason about a single question across the whole horizon: can this demand be met, and if so, exactly what must happen and when.

The planning cycle begins with demand review, where planners inspect and adjust the forecast, confirm how firm orders consume it, and settle the demand plan that will drive the period. From there the master schedule is set or revised, either manually for stability or by the planning engine within defined tolerances, with the time fence protecting near-term commitments from disruptive change while allowing the far horizon to flex.

The core workflow is the planning run. A regenerative or net-change run explodes requirements, nets supply against demand, and produces planned orders together with a full exception list: orders to release, orders to reschedule in or out, projected shortages, and past-due conditions. Planners work this exception list rather than the raw plan, acting where the calculation flags a problem. Recommended actions can be firmed and released as work orders to the shop floor or as purchase requisitions to procurement in a controlled step, with automatic release available for items whose policies permit it.

Capacity resolution runs alongside. Where the finite schedule reveals an overload, planners rebalance by moving work between resources, adjusting the schedule, expediting supply or authorising overtime, and re-run to confirm the plan is once again feasible. Available-to-promise and capable-to-promise queries let order management commit realistic dates back to customers directly from the live plan.

The module implements the full set of lot-sizing techniques planners rely on, from lot-for-lot and fixed order quantity through periods of supply, economic order quantity and min-max, each configurable per item and honoured through every explosion. Lead-time management is layered, separating fixed and variable, run and queue and safety lead time so that planned dates reflect how the item genuinely flows. Safety stock and safety time are both supported, and dynamic policies let buffers respond to demand variability rather than sitting static.

Forecasting carries a genuine statistical toolkit, including moving average, exponential smoothing with trend and seasonality and regression, with forecast accuracy tracked continuously through error measures so that models are held to account and re-tuned. Pegging is maintained end to end: any requirement can be traced upward to the specific demand that created it and downward to the supply that covers it, which makes the impact of a single late order or a single design change fully visible. The engine performs both regenerative and net-change planning, so a small change updates only what it touches. Finite scheduling honours forward, backward and bottleneck-first sequencing, resource calendars, changeover and sequence-dependent setup, and constraint-based loading, so that the schedule it produces is one the plant can actually run. Throughout, the plan is valued, letting planners weigh a feasible schedule against its inventory and capacity cost.

Production Planning is defined by its connections. It draws product structures and routings directly from Manufacturing and Operations, so bills of material and work-centre definitions are never re-keyed and always current; the work orders it releases execute there, and completions and consumption flow back to keep the plan honest. It hands planned purchases to Supply Chain as requisitions with correct dates and quantities, and it reads open purchase orders and supplier lead times back so that inbound supply is netted accurately.

Demand originates in the order and forecast signals raised through Sales and CRM, and the plan's promise dates feed straight back into order commitment. Inventory positions are shared with the stock and warehouse capabilities that hold on-hand and in-transit balances. Financials consumes the valued plan for standard costing, inventory valuation and budgeted purchasing, while Human Capital and resource scheduling inform the labour and shift calendars that bound finite capacity. The result is one plan, calculated once, that the whole enterprise reads from a single source rather than reconciling between disconnected spreadsheets.

AI Workers act as first-class participants in the planning process, not as a bolt-on. A planner can ask, in plain language, why a component is short next week, and the Worker traces the pegging, names the demand that drove it and the supply that failed to cover it, and explains the gap. Workers execute actions directly within the guard rails set for them: firming and releasing recommended orders, rescheduling within tolerance, and re-running net-change planning after a change, each action logged and attributable.

They watch the plan continuously for the conditions humans would otherwise chase. A Worker raises anomaly and exception alerts when forecast error drifts, when a work centre tips into sustained overload, when safety stock is breached or when a supplier lead time moves in a way that puts the master schedule at risk. Workers extract structure from unstructured input, reading a supplier confirmation or an engineering change notice and turning it into a proposed plan adjustment for review. In decision support they compare feasible schedules, quantify the inventory and capacity trade-offs of each option, and recommend a course. And they stand as named nodes in approval and review workflows, signing off routine order releases within policy and escalating the exceptions that genuinely need a human, so the plan moves at machine speed where it is safe and human judgement is reserved for where it counts.