MCP for production
Investigate downtime and process data with targeted queries
When a line stops, clues are often spread across controllers, control systems and shift logs. We connect approved sources so maintenance and shift supervisors can investigate relationships. Engineering for production data has its own requirements for plant load, network segmentation and acceptance.
The connection
The standard route is OPC UA, because it is available in most modern plants and comes with a clean information model. Where that is not enough, Modbus, MQTT or a direct route to the controller are added, for example with Siemens S7 or Beckhoff. On the analysis side, we connect MES and control systems so that the order and batch context is preserved.
The architecture preserves the separation between office and production networks. Access uses approved interfaces and defined network boundaries; the plant level is not directly accessible to a model. Which results an AI application or model provider receives is agreed independently of where the MCP server is hosted.
Reading is the normal case
Most of the benefit arises without a single write access: downtime analysis across several shifts, development of overall equipment effectiveness, changeover times compared between lines, quality data in relation to the batch. Questions that nobody asks today because the analysis takes longer than the benefit is worth.
- Fault analysis
- “What happened on line 3 yesterday between 2 and 3 pm?” — with the controller messages, the preceding process values and the order context in one answer.
- OEE and availability
- Development over weeks, broken down by cause group, without anyone building a new report.
- Changeover and retooling times
- Comparison between shifts, products and plants, as a basis for concrete improvements rather than assumptions.
- Quality and traceability
- Process values in relation to the batch, when a complaint comes in and the question is what distinguishes this batch from the others.
Control: possible, but not in passing
Write functions are a separate engineering commission with risk assessment and their own acceptance process. The following controls are minimum requirements and do not replace assessment of the specific plant or its safety design.
- Allowlist instead of general authority
- Only explicitly approved operations can be called at all, with fixed value ranges. Whatever is not on the list does not exist for the model.
- Confirmation by a human
- Every write operation is displayed and must be approved by an authorised person. The model proposes; it does not act alone.
- Complete audit log
- Who, when, which operation, which values, which approval. Traceable and analysable, independently of the MCP server.
Why someone from engineering should build this
The difference between a connection that works in the lab and one that can hold up in running production does not lie in the AI. It lies in the question of which cycle times a query may tolerate, how a connection loss behaves and what happens if someone switches the plant to manual mode in the meantime.
Schady Hamid, B.Eng, leads production and engineering. Implementation takes plant states, network segmentation and operational ownership into account. Load and failure tests, together with a separate approval, are part of preparing for production operation.
Limits we state up front
- No safety-related functions. Emergency stop, safety circuits and everything under functional safety remain outside the scope.
- No automatic intervention without human approval.
- No removal of the network separation between office and production networks.
- No connection to plants whose manufacturer does not approve the interface.
Frequently asked questions
Can the AI actually control our machines?
Selected operations that are not safety-related may be technically possible. Their suitability is assessed in a separate engineering commission and accepted separately. Minimum requirements include fixed value ranges, an allowlist, human approval and an audit log. Emergency stop and other safety-related functions remain excluded.
Which protocols and controllers are supported?
The standard route is OPC UA. In addition, Modbus and MQTT as well as direct connections to common controllers such as Siemens S7 and Beckhoff. On the analysis side, we connect MES and control systems so that the order and batch context is preserved.
Does this put load on our plant?
Queries create load. We limit data volume and query rates and use suitable interfaces or historian data. Acceptable load limits are agreed with your IT/OT and maintenance teams, and effects on plant operation are tested before production approval.
Does the production network have to be opened for this?
Network separation remains in place. Your IT/OT owners review and approve the connections needed between the interface, MCP server and AI application. Plants are not exposed directly to the model provider; selected query results may be transmitted to it, depending on the approved architecture.
Is it worthwhile even without an MES?
It may be worthwhile if suitable controller and process data are available for the agreed question. To make statements about stoppages, changeover times or availability, we check data quality and time references. An MES can add order and batch context.
Let’s check whether it holds up for your systems.
We first clarify your use case, data sources, permissions and acceptance criteria. The feasibility review documents prerequisites and limits before a pilot is agreed.
Discuss your use case