Marc Krüger-Sprengel
Legacy Lines, Real Pressures
Decision intelligence for manufacturing has one hard prerequisite: an architecture built for the systems you already run. Most factories don’t begin with a clean sheet. They evolve over decades, layer by layer. PLCs from the 1990s still run beside newer CNCs. MES implementations remain half-finished. Excel bridges process gaps that no software ever closed. Each site has its own naming rules, integration methods, and long-standing workarounds known only to experienced staff.
This is the operating reality of modern manufacturing. It keeps production running, but it also limits progress. Every digital initiative starts with the same promise to “clean up the data,” and months later, teams are still cleaning while project goals slip.
Replacing legacy systems isn’t realistic. Connecting them intelligently is.
What “Context First” Means in Practice
A Context Layer unifies what manufacturers already have: assets, processes, materials, quality records, and documents. It brings together information from MES, ERP, historians, and controllers into one dynamic model: a Unified Context Graph.
This model doesn’t demand a perfect schema. It learns the plant as it operates and continues to evolve as the work changes. When a new process step appears or an operator logs a deviation, that information becomes part of the graph automatically.
A context-first approach reverses traditional data design. Instead of forcing reality into rigid models, it lets structure emerge from live production. This makes it fit for brownfield plants, where variety is the rule, not the exception.
How It Works
Connectors capture live data from OT, IT, and quality systems via OPC UA, MQTT, REST, or direct database access without disrupting production.
Translation layer uses industrial foundation models, supported by engineers, to map tags, tables, and documents into shared semantics (asset - process - material). Late binding ensures flexibility; lineage ensures trust.
Explorers and APIs provide role-based views for operations, engineering, and planning, and allow internal or vendor applications to consume the same context without redundant integrations.
Together, these elements form a live operational backbone that reflects the real state of production and enables improvement without downtime.
Design Principles for Brownfield Deployment
Brownfield by default: Work with existing infrastructure instead of forcing rewiring.
Schema evolves with use: Standardization follows proven value, not theoretical design.
Governance in the fabric: Access control, approval flows, and data lineage are embedded, not added later.
Edge-aware: Run workloads near the line when latency matters, then synchronize to the cloud for scale and analytics.
This approach replaces multi-year replatforming with incremental, measurable progress. Start on one line, validate the benefit, and expand.
Why Scaling Fails Without It
According to McKinsey, 74 percent of Industry 4.0 programs fail to scale beyond pilots. The reason isn’t lack of ambition, but lack of scalable architecture. Integration costs rise sharply, and attempts to enforce rigid standards collapse under the diversity of real plants.
Brownfield facilities are the norm, not the edge case. The true constraint is connecting them. A context-first layer creates a scalable backbone that works with heterogeneity instead of resisting it.
What Users Experience
Operators access a unified view that combines traveler information, sensor data, and context. They no longer need to switch between MES, historians, and document folders to diagnose a problem.
Quality leads can trace a defect through materials, equipment, and process parameters within seconds.
Planners can simulate “what-if” capacity changes, such as tool replacements or downtime, based on current data instead of static spreadsheets.
All teams use the same live model, ensuring that insights and actions align.
Why MES and Data Lakes Fall Behind
Both MES and data lakes were built for a world that changed slowly.
Manufacturing Execution Systems (MES) were designed to enforce consistency in high-volume production. They track orders, enforce workflows, and minimize deviation. That works when products are stable and lines rarely change. In modern factories, product variants shift weekly and production adapts continuously. MES systems assume uniformity; brownfield operations depend on flexibility. The result is longer change cycles, parallel spreadsheets, and delayed improvement. MES becomes the very bottleneck it was meant to remove.
Data lakes were meant to bring flexibility, but they treat industrial data like office data: static, disconnected, and context-free. Lakes store information efficiently, but strip it of relationships: how a sensor tag links to a batch, how that batch connects to a process, how the process impacts yield. The data exists, but meaning is lost.
A Context Layer bridges these gaps. It keeps MES for execution and compliance but allows it to interact with other systems through shared context. It keeps the data lake for storage but adds the semantic structure required for analytics, AI, and continuous improvement.
MES enforces, data lakes collect, context connects and explains.
Brownfield as a Strategic Advantage
Once context becomes accessible, diversity turns into a strength. Sites can share proven improvements without forcing identical setups. Local optimizations remain intact, yet become transparent and comparable. This allows networks to learn and scale without cloning entire plants.
Manufacturers that lead this shift are focusing less on new tools and more on adaptable architecture, connecting what already works while preparing for what comes next.
Build the Stack Behind the Stack
Waiting for a “greenfield moment” means standing still. The next competitive advantage lies in architecture that embraces the systems you already run.
context/fab deploys decision intelligence for manufacturing that connects OT and IT, enriches data with meaning, and delivers measurable results quickly, without replacing systems or pausing production. Start with one line, then scale the most impactful use cases across your manufacturing network.