Order dressing
Translate commercial orders into producible production orders — grade mapping, dimension tolerances, over/under delivery rules, route selection.
Home / DigiPlantMES
Our product
DigiPlantMES — a Manufacturing Execution System built for metals, not adapted to it.
Modular. Multi-tenant. Deployed per customer as a versioned unit. Covering steel, aluminium, copper and special alloys from charge to despatch — on a PostgreSQL stack with no database licence attached.
Metals covered
Where it sits
Level 3, and honest about it
DigiPlantMES owns the execution layer: what is being made right now, from what, to which specification, and what happened to it. It does not pretend to be your ERP, and it does not try to replace your L2 process models.
That boundary is deliberate. The programmes that go badly are usually the ones where an MES quietly absorbed order management or setup calculation because it was easier than integrating properly. We integrate properly.
- Receives orders, contracts and material masters from ERP
- Sends confirmations, consumption, yield and cost-relevant events back
- Exchanges setup and actuals with L2 over OPC-UA, MQTT, Modbus or file drop
- Publishes a clean event stream for analytics without becoming the warehouse
ISA-95 positioning
Forward and reverse traceability in one model
The core
Material genealogy that survives the audit
Every piece of material carries its full ancestry — charge lots, heat, chemistry, cast, reheat, rolling, coating, cut — and every quality record hangs off the node it belongs to.
When a customer calls in fourteen months' time about a coil that failed a bend test, you need to know which heat it came from, which other coils shared that heat, where they went, and whether the same caster sequence produced anything else at risk. In DigiPlantMES that is one query, not a week of spreadsheets.
Forward trace
From a raw material lot to every finished piece and customer it reached.
Reverse trace
From a complaint back to charge composition, process actuals and operator.
Split & merge
Cuts, re-coils, re-melts and blends modelled properly — not as a comment field.
Certificates
Mill test certificates generated from the same data, not re-keyed from it.
Modules
Nineteen modules. Take the ones you need.
Modular is not a marketing word here — modules are separately licensable and separately deployable. Start with tracking and genealogy on one line if that is all the appetite you have.
Material tracking
Real-time position and state of every heat, slab, billet, coil, plate and bundle across the route, driven by L2 events and manual confirmations.
Material genealogy
The full ancestry model — splits, merges, re-melts and re-work, with every quality record attached to the right node.
Quality & specifications
Internal specs, customer specs, test plans, sampling, dispositioning and non-conformance handling with proper approval trails.
Business rules engine
Thirteen rule models covering practice selection, routing, disposition and alerting — configured by your engineers, not by a change request.
Yard & 3D visualisation
Stockyard layout, stacking rules, coil pyramids, crane moves and pick strategy — rendered in 3D so the constraint is visible, not implied.
Integration layer
Kafka and Celery under the hood, with configurable XML/JSON field mapping, replay and a readable dead-letter queue.
Tag library
OPC-UA, EtherNet/IP and Modbus tag definitions managed centrally, so a new gauge is a configuration change, not a project.
Shift & crew management
Shift patterns, crew assignment, handover notes and shift reporting that ties production back to the people who made it.
Reporting & ATV
Aggregated transposed views give planners fast, wide reporting over deep production data without hand-written SQL.
FerroGPT assistant
A retrieval assistant over your own production data — 370 pre-built queries across nine sections, so a planner can ask instead of filtering.
Scheduling modules
Melt shop, caster, hot mill, cold mill, finishing and forging schedulers. See the suite
Under the hood
Architecture choices we will defend
Every one of these was made against a ten-year plant lifecycle, not a demo.
Modular monolith, not microservices
A service-based architecture deployed as one versioned unit per customer. A mid-size mill should not need a platform team to run their MES — and a broken service mesh at 03:00 is not a fair trade for architectural fashion.
PostgreSQL, all the way down
215+ tables across 19 apps, with no Oracle licence in the total cost. For a mid-market producer this is often the single line item that moves a project from "impossible" to "approvable".
A grid that survives real data
Our MES-GRID renders 50,000 rows in under two seconds and is embedded inside React rather than replaced by it. Planners live in grids; the grid has to be quick or the system is not used.
Schema management done properly
Liquibase layered over Django migrations, so a customer three releases behind can still be upgraded predictably instead of manually.
Multi-tenant, single-tenant deployable
Multi-tenant in the data model, but deployable per customer on your own infrastructure or in your own cloud tenancy. Plant data does not have to leave your boundary.
Release trains, not big bangs
Semantic versioning, a release-train branching model and per-customer release management — so upgrades are routine rather than an annual crisis.
Positioning
Where DigiPlantMES fits — and where it does not
We would rather tell you this now than in month nine of an implementation.
A good fit if
- You are a mid-size producer — roughly 0.3 to 3 Mt/year
- Tier-1 platform pricing has already ruled those options out
- You want to start with one area and extend, not sign a five-year programme
- You have engineers who want to configure rules themselves
- Data residency or on-premise deployment matters to you
Probably not, if
- You need a fully proven install base of 50+ reference plants today
- You want a single vendor to also own L2 automation and drives
- Your requirement is really an ERP or an APS, not an MES
- You need it live across six plants in six months
Design Partner Program · open
Three plants will shape this product. Then the terms change.
We are selecting a small number of design partners to build DigiPlantMES against real operations rather than assumptions. You get the product on terms that will not exist again; we get a reference and a reality check.
Up to 100% licence discount
For the first three customers, scaled to the reference commitment you make.
Five years of free upgrades
Every release on the train, at no additional licence cost.
Source code escrow
Continuity protection, so a mid-size supplier is never a single point of failure.