INFOLINK MES
INFOLINK is a manufacturing execution system (MES).
Traceability
Trace any finished product back to the material that went into it.
Control
A stage cannot begin until the one before it is complete.
WHY INFOLINK
Three situations that come up on floors running without an MES, and how INFOLINK answers each.
A customer reports a defect. Answering which materials were used and when means digging through paper logs and staff recollection. Unable to bound the recall, teams cast the net wider than necessary.
INFOLINK
A LOT (a batch of material produced under the same conditions) number is assigned at goods receipt and carried through inspection and production to the finished product. The link between material LOT and production LOT is written automatically as output is reported, so nothing has to be compiled after the fact.
The month-end count disagrees with the ledger. With no way to find where it diverged, the number simply gets adjusted.
INFOLINK
Every action that changes stock passes through a single function, and that function updates the balance and writes the ledger entry in the same transaction. Ledger total therefore always equals current stock, and when it does diverge the responsible transaction can be identified.
When inspection results live outside the system in spreadsheets and on paper, there is no way to stop rejected material from entering production. A written rule gets skipped on a busy day.
INFOLINK
Material that has not passed inspection is excluded from the issue-candidate list outright. It is not blocked by policy; it does not appear in the results.
PROCESS
This is not a list of screens. Each stage feeds the next, and work cannot proceed unless the conditions are met.
Sales
Order intake
Purchasing
Purchase order
Goods receipt
LOT assigned
Inspection
Disposition
Production
Work order · output
Quality
In-process · final
Shipping
Dispatch
Closing
Ledger close
Which module covers which of the eight stages.
MRP (material requirements planning — what to buy, when, and how much, derived from the production plan) review sits within production planning; output reporting sits within production.
It is not that the step can be skipped and should not be. It cannot be taken unless the conditions are met.
That constraint is what enforces quality.
HOW IT WORKS
What operators enter on the floor becomes the management screen itself. Nothing is transcribed or re-totalled.
DATA COLLECTION
Output, inspection and stock movements are recorded where they happen. Operators enter them at the POP shop-floor terminal, and what goes in is organized by process and becomes the input to the next stage.
MONITORING
Output, defect rates and stock on one screen. The shop floor terminal and the executive dashboard read the same data, so the meeting spent reconciling numbers goes away.
FEATURES
What each one solves comes first; the detail and the evidence sit underneath.
Answer which material was used and when, the moment a defect is reported.
Evidence
A dedicated genealogy table links material LOTs to production LOTs, written automatically as output is reported.
Make it possible to retrace why book stock and physical stock differ.
Evidence
Every action that changes stock passes through a single function, which updates the balance and writes the ledger entry in the same transaction. Ledger total therefore always equals current stock.
Accept or reject against registered criteria rather than recollection.
Evidence
The acceptance criteria are read from the server, not taken from the values the screen submits.
One continuous flow from work order to completion, with no step quietly skipped.
Evidence
BOM and routing are copied and frozen at the moment the work order is issued. A later BOM change does not alter the requirements of an order already released.
Splitting and merging LOTs on the floor never happens off the record.
Evidence
A cancellation does not erase the record. A reversal writes a new opposing entry, so what was cancelled and why remains visible.
Help for when the data is there but it is unclear where to look first.
The AI does not decide for you. It points at where to look first. The judgement and the action stay with your team.
Evidence
Statistics and aggregation are done by the system; the AI explains the result. It does not fix a cause, only proposes candidates and an investigation order, and it does not reorder that ranking. It is read and analyse only, and takes no automatic action such as stopping equipment, holding stock or raising a non-conformance.
TECHNOLOGY
What is implemented today.
One database: PostgreSQL
No burden of operating several stores side by side.
No front-end build step
Change a file and it takes effect. Deployment stays simple.
Light on people
Light to adopt, and light to keep running.
Start Small does not mean features were taken out. It means the system was built this way.
Permissions are not a matter of hiding screens. The server checks every request as it arrives, so revoking a permission takes effect immediately.
ARCHITECTURE
A layered structure running from the user-facing screens down to the data layer. What is implemented today is set out below.
The diagram above shows the architecture INFOLINK is built toward. The current implementation centres on a PostgreSQL operational database and a FastAPI service layer; time-series and analytical stores and container-based deployment are not part of it yet.
A data-driven manufacturing operations platform
Scroll sideways to see the full diagram
Executives, managers, supervisors, operators, quality staff, suppliers and customers each get their own view — on the web, a mobile app, a shop-floor PDA or scanner, or a dashboard.
Eight modules: sales and orders, production planning, purchasing and materials, work in process, quality, equipment, shipping and logistics, tracking and analytics. Workflow, approvals, notifications, permissions, audit logs and localization are shared services, carried by an API gateway and master data services underneath.
Quality deviation analysis, analysis screens for equipment, inventory and planning, analysis output as documents (Excel, Word, PDF), and analysis history kept for audit. Read and analyse only, with no automatic action.
Operational database (PostgreSQL), time series (InfluxDB), warehouse (DuckDB) and file storage (MinIO/NAS/S3), each for its own purpose, connected by streaming and batch ingestion pipelines.
Runs on-premise or in a private cloud. Docker and Kubernetes deployment, Prometheus, Grafana and ELK monitoring, backup and DR, firewall, SSL and VPN.
Connects to ERP (finance, purchasing, HR), PLM (design, BOM), SCM and customer systems over EDI.
One set of numbers
The shop-floor terminal and the executive dashboard read the same source
A record that stays
Totals and history remain as the basis for a decision
A fixed order
A stage opens only when the one before it is done
Steady disposition
Judged against registered criteria, with the history kept
Stepwise growth
Adopt module by module and widen when you need to
Just tell us what you make and how you track it today. We will map out what fits and reply within one business day.
