ERPNext's native Production Plan and Work Order engine is excellent at answering "what and how much to make" — it explodes BOMs, nets against stock, and raises material requests. What it does not do is answer "when, on which workstation, in what sequence, and can we actually hit the promised date." Native scheduling assumes infinite capacity: two Work Orders needing the same CNC on the same shift are both happily "planned," and the Gantt in the desk is a static read-out, not a plan you can push, pull, and re-optimize. For a discrete or job-shop manufacturer running dozens of concurrent orders across shared workstations and skilled operators, that gap turns into missed due dates, expedite chaos, and planners maintaining the real schedule in a spreadsheet outside ERPNext.

We build a proper Frappe app — its own installable module, not a pile of loose customizations — that layers finite-capacity, constraint-aware scheduling on top of your existing ERPNext manufacturing data. The core is a new APS Schedule DocType plus per-operation APS Task child records that map one-to-one onto your Work Order operations and Job Cards, so nothing is duplicated or forked away from standard ERPNext. A whitelisted scheduling service (exposed via frappe.whitelist() and the standard REST API) runs a finite-capacity pass over each Workstation's real working-hours calendar and operator availability, honoring setup time, run time per unit, and material readiness dates pulled from stock projected-qty and expected PO/MR receipts. The result is written back to Job Card from_time/to_time and Work Order planned_start_date, so downstream ERPNext (capacity reports, Job Card execution, timesheets) sees a schedule it already understands.

The planner works in a drag-and-drop Gantt board delivered as a Frappe client script / custom page: drag an operation and the app auto-reschedules dependent and competing tasks, flags any capacity or material conflict in real time, and refuses to silently double-book a workstation. Sequencing rules are configurable — earliest due date, priority weighting, and setup-time minimization (grouping like setups to cut changeovers) — and a what-if mode lets a planner clone the live schedule into a sandbox, simulate a rush order or a machine breakdown, see the bottleneck workstation and the knock-on late orders, then commit or discard. Server-side hooks.py doc events keep the schedule honest as reality changes (a Work Order is submitted, a Job Card completes, a Material Request is received), and a scheduler_events background job can run a nightly full re-optimization so each morning starts from a feasible plan. Role profiles and permissions gate who can plan versus who can only view, integrated with your existing ERPNext manufacturing roles.

Because this is build-to-order, we start from your actual routings, workstations, shift calendars, and scheduling rules rather than a generic template. After a scoping call we confirm the exact DocType fields, sequencing logic, and integration points, then build against Frappe/ERPNext v15 (and v16 where you are already on it), validate on a staging clone of your site under UAT, and hand over the full git repository. Typical delivery is 2–4 weeks from confirmed scope. There is no instant download — you get a version built for your plant, installed with bench get-app/bench install-app, documented, and supported.