← back

Untitled

Fine. Now ask the impolite question.

Can the factory establish what happened, where it happened, when it happened, and whose authority made it acceptable?

If the answer arrives wrapped in a presentation, the transformation has already failed its first serious test.

A connected line can stream temperatures, pressures, alarms, recipes, and operator actions while leaving the meaning of those events fractured across a historian, a manufacturing execution system, a laboratory system, an electronic batch record, a quality system, and several spreadsheets that nobody wants to admit still matter. Rockwell’s report says industrial data operations should connect, contextualize, and govern operational data across equipment, processes, and manufacturing systems. It also calls for standardized metadata, models, lineage, and access.

That recommendation deserves more attention than the headline percentage.

The collision sits between strategy and systems. Executives announce transformation while engineers and quality teams spend hours proving that “Line 3,” “Mixer 04,” and “Asset 0004” refer to the same physical object. A material code changes after a supplier qualification. A batch exists under one identifier in the planning system and another in the laboratory record. A deviation points to a record whose authority is stored somewhere else. The release decision arrives with a digital signature, yet the surrounding evidence cannot always explain which version of the process was actually executed.

The screen glows. The meaning is missing.

Picture a quality investigator at 2:17 in the morning. A deviation has triggered a review before release. She has the alarm history, though the timestamps use a different convention from the batch record. She has a laboratory result, though its sample identifier does not resolve cleanly to the equipment run. She has an approval trail, though the system shows a role rather than a person with authority over that specific decision. Nobody needs another dashboard. She needs a defensible chain of identity.

That chain is where pharmaceutical software becomes consequential.

Equipment needs a durable identity across the asset registry, control layer, maintenance system, historian, and manufacturing execution system. Materials need governed relationships to suppliers, lots, specifications, substitutions, and test results. Batches need lineage that survives transfers between planning, execution, laboratory, and quality systems. Deviations need links to the exact event, version, action, and approval that shaped their disposition. Release records need evidence that can be assembled with context rather than excavated from disconnected applications.

Rockwell’s proposed evidence layer points in the right direction. The report describes secure portals, curated exports, standardized record packages, and audit trails capable of assembling batch records, deviations, corrective and preventive actions, change controls, validation evidence, and product quality reviews. That is a serious operating model. It also quietly admits the uncomfortable truth: readiness depends on the ability to retrieve meaning, not merely data.

Manufacturers often stall because identity work feels less glamorous than installing a new platform. Engineers know the cost immediately. Someone must reconcile naming conventions, define master data ownership, map interfaces, preserve historical context, test exception paths, and decide what happens when two systems confidently disagree. Quality teams must participate early because a semantic decision can alter the interpretation of a record. Security teams must govern access at the boundary where operational technology meets enterprise applications.

The work is slow because it is real.

A modern data platform cannot repair an undefined relationship between a recipe version and the batch that consumed it. An application programming interface can move a deviation record while stripping the event context that makes the deviation intelligible. A digital signature can prove that someone approved a decision while leaving the authority model too vague to defend why that person was permitted to approve it.

This is why “connected factory” has become such a dangerous phrase. It encourages leaders to count connections instead of asking whether those connections preserve identity, provenance, authority, and intent. A factory may have hundreds of integrations and still force a reviewer to perform archaeology.

The report’s focus on cyber resilience sharpens the point. Rockwell says resilience must extend from connected products through design, deployment, and maintenance, with vulnerability management, governance, and supply chain risk handled as part of trusted operations. Security boundaries now shape evidence boundaries. If an operator identity, equipment state, or change record cannot be trusted across that boundary, the quality argument weakens with it.

Continuous readiness therefore means more than being prepared for an inspection. It means the plant can explain itself while work is happening. A release decision should be supported by evidence that retains its relationships. A deviation review should expose the affected process state without sending investigators on a tour of five systems. A change control should reveal which equipment, materials, procedures, and records fell within its reach.

That outcome requires engineering discipline. Define canonical identifiers. Preserve source identifiers alongside them. Record lineage. Version the models. Treat metadata as part of the product. Give quality a seat in the data architecture instead of inviting the department after the migration has been blessed. Build user interfaces that show provenance and authority in the same working context, because forcing a reviewer to assemble those facts mentally is a human factors failure disguised as flexibility.

The industry has been selling unfinished meaning as momentum. Rockwell’s 90 percent figure makes the sale louder, while its own recommendations reveal the plumbing still exposed beneath the floorboards.

Senior leaders should challenge the premise that digital transformation is automatically business critical. Transformation earns that status when it can establish what happened, where, and under whose authority, then produce the evidence without theatrical reconstruction. Until then, the connected factory is a beautiful stage set with a missing script.

Pharma does not need another slide showing glowing machines. It needs records that can survive scrutiny, systems that can explain their relationships, and software designed around the moment a batch must be released with confidence.

If this handoff problem is sitting inside your plant, hello@example.com is a sensible place to put the first uncomfortable question.' tags: - weekly-hype - semantic-data - interoperability title: The Factory Is Connected. The Evidence Is Still Missing. type: weekly_hype

Related

Federated Scientific Data Has to Move Decisions, Not Decorate ScreensThe Plumbing That Breaks When ELN/LIMS Data ScalesThe API Layer That Crosses Silos Without Moving Data