Enterprise: ERP, CRM and SAP · Enterprise scenario · Proposed architecture
From sales forecast to production plan (Demand-to-Production)
From sales history, open orders and the CRM pipeline to an approved production plan: a forecast model calculates scenarios, the planner approves the demand plan, raw materials, capacity and supplier lead time are checked in parallel, and execution documents are created once in the ERP.
Who it is for: Manufacturers and businesses with seasonal demand
Systems: CRM, ERP, SAP production and material planning processes, Warehouse system, Forecast model
Problem
The sales forecast lives in Excel, the pipeline in the CRM, and material and capacity data in the ERP. The planner reads campaign notes separately, and raw material shortages surface only after the plan is approved.
How it works today
At month end the planner collects sales and order data by hand and calculates the forecast in a spreadsheet. Material requirements are worked out separately, alignment with procurement and production happens over email, and orders are entered into the ERP by hand.
What this scenario changes
The planning cycle collects data automatically and checks its completeness. A separate forecast model calculates scenarios, and AI summarizes campaign notes. The planner approves the demand plan, material requirements are calculated, and raw materials, capacity and lead time are checked in parallel. On a shortage an alternative plan is prepared, and after procurement and production approval the execution documents are created in the ERP with an idempotent write.
The integration method is chosen based on the SAP version and the available interfaces.
Process stages
- Data collection: Gather sales, order and pipeline data in one place. Result: Initial data set for the planning cycle
- Data quality: Check completeness and consistency across periods. Result: Data fit for forecasting. Approval: Demand planner (on gaps)
- Forecast scenarios: Model calculation and campaign context. Result: Base, high and low demand scenarios, plus a summary of notes
- Demand plan approval: The planner chooses and approves a scenario. Result: Demand plan with a recorded version. Approval: Demand planner
- Materials and constraints: Material requirements, raw materials, capacity and lead time. Result: Combined result of the three checks
- Alternative plan and approval: An alternative on a shortage and approval of the execution plan. Result: Plan approved by procurement and production. Approval: Procurement and production leads
- Execution documents in the ERP: Create the documents exactly once. Result: Planned Order and Purchase Requisition
- Tracking actual execution: See the gap between planned and actual production. Result: The variance report is passed to the next cycle
Participants
- Demand planner
- CRM
- Forecast model
- Integration layer
- ERP / SAP
- Procurement and production leads
- Shop floor and warehouse
The role of AI
Summarizes campaign and market notes for the planner and explains the differences between plan scenarios. The numeric forecast is done by a separate model, and planning follows ERP rules.
Rule-based automation
Data completeness, material requirements and constraint checks run on rules. Planned order and purchase requisition writes are idempotent through an operation ID, and on a timeout the ERP is checked first.
Human approval and exceptions
The planner approves the demand plan, and the procurement and production leads approve the execution plan. If demand changes after approval (for example, a large order is cancelled), the earlier approval is not reused. When the retry limit is reached, an integration operator handles recovery.
Frequently asked questions
Do you have a ready-made SAP integration?
This page shows the proposed architecture and a synthetic demo. The integration method is chosen in the project based on the SAP version, the active production and planning processes, and the available interfaces.
Does AI calculate the forecast?
No. The numeric forecast is calculated by a separate model, and material requirements by ERP rules. AI summarizes campaign notes and explains the differences between plan scenarios, and people approve the plan.
What happens if a large order is cancelled after approval?
The approval is tied to a specific plan version. If demand changes, the plan goes back to the planner before any documents are created, and materials and constraints are checked again.