A commissioning team solves an unexpected problem. A control routine stabilises production, a design change reduces maintenance effort, or a diagnostic method helps an older machine perform more reliably. The customer receives a working solution and the project moves forward. The same development may also offer value for other plants, future retrofits or a digital service. Recognising that wider potential creates an opportunity to turn everyday engineering into a stronger technology base.

This opportunity is particularly relevant in machinery and plant engineering, where performance depends on the interaction of physical equipment, automation software and accumulated process experience. The new CEIPI lecture, AI Supported IP Management in Industrial Engineering, explores how companies can identify valuable project knowledge, make informed protection decisions and prepare it for reuse. Its central concern is a process that engineers can use within the demands of project delivery.

The decisions hidden inside project work

A technical improvement creates several connected questions. What customer outcome does it improve? Could other projects benefit? Which part belongs to a reusable company platform, and which part reflects confidential customer conditions? Does a supplier contribution affect the company’s freedom to use the result? Answers influence both the protection strategy and the economics of future engineering work.

Consider a hypothetical improvement to strip-flatness control. Its value could depend on a sensor arrangement, a control algorithm and calibration experience. Copying the code into another project may leave essential operating knowledge behind. A patent application may address a technical contribution, while confidentiality measures and agreed permissions remain relevant to other parts of the solution. A useful IP process considers those dependencies together.

Why established management theory helps

David Teece’s work on profiting from innovation explains why returns depend on the conditions for protection and the complementary capabilities needed to commercialise a development. In industrial engineering, these capabilities can include commissioning expertise, integration skills and access to installed equipment. They help explain why the business value of a solution extends across several technical and organisational layers.

Ikujiro Nonaka’s theory of organisational knowledge creation draws attention to the relationship between documented knowledge and practical experience. Reuse needs enough context for another team to apply a solution successfully. Cohen and Levinthal’s concept of absorptive capacity adds a perspective on external knowledge: prior expertise helps a company recognise and apply valuable inputs from suppliers, research and technology intelligence. These theories give managers concrete reasons to connect information processing with expert judgement.

A process that fits the engineering workflow

The lecture follows five stages: Identify, Evaluate, Decide, Protect and Reuse. AI can help search authorised project records, cluster related developments and prepare an initial assessment. Every candidate should remain connected to its evidence, contributors and project context. Engineers can then validate the technical finding, while business owners assess its wider relevance and IP experts examine the legal questions.

The evaluation should distinguish commercial priority from legal assessment. A promising reuse opportunity may justify documenting know-how even when a patent route looks uncertain. An imminent disclosure may require urgent review despite a modest initial priority. The process needs an owner for each decision, a record of the reasons and a route for resolving uncertainty. Approved versions and permitted uses then make the outcome available to subsequent projects.

Siemens shows how AI can enter engineering work

Siemens reported in October 2024 that more than 100 companies were using its Industrial Copilot. Its thyssenkrupp Automation Engineering example concerns battery-machine development, including sensor configuration and documentation. Siemens also reported generated code requiring only 20 percent adaptation. That figure describes a vendor-reported capability, with no published evaluation protocol in the announcement. It provides no measurement of invention identification or patent quality.

The knowledge boundary is equally instructive. In its 2023 announcement, Siemens stated that customers retained control of their data and that it would not train the underlying AI models. For an IP process, the transferable design principle is to combine useful assistance inside engineering workflows with explicit rules for proprietary information. The next step is to test whether similar support produces better candidate evidence and more timely decisions.

Bosch Rexroth connects modularity with commercial permissions

Bosch Rexroth’s ctrlX AUTOMATION offers a complementary example. Its software development kit enables users to integrate existing code and their own know-how as applications. The developer guideline describes app licensing and validation requirements. This gives software providers a structured route for making functions available through an interoperable industrial platform.

An implemented retrofit at the Tillmann Group, reported in August 2026, illustrates the operational setting. Bosch Rexroth modernised a metal-profiling machine while retaining its higher-level third-party control system. The company reported 30 percent less control-cabinet space and 25 percent lower energy use. Those results concern the complete retrofit, including drive-system changes. They do not isolate the contribution of software licensing or IP management.

Together, the platform documentation and the retrofit show how reusable functions can coexist with integration into an established production environment. The lesson for project businesses concerns defining the interface, the reusable implementation and the permissions attached to it. The two industrial examples support different parts of the lecture’s framework. Neither documents the complete five-stage AI-supported IP process.

From the lecture to the expert application

The slides connect these examples to decisions about customer adaptations, joint improvements and operational learning. The accompanying platform reading extends the discussion through AI in Operational IP Management, Know-how Management and the Research Nugget When AI Enters Patent Practice. Together, they help readers assess where automation is useful and where professional judgement carries the decision.

The planned next step is an expert model solution for Max Feucker addressing the associated practical question. No model solution is available yet. Its intended contribution is to connect candidate selection, protection choices and rights-cleared reuse in a workable process, explaining the trade-offs for everyday project business. The lecture below provides the foundations and industrial comparisons for exploring those decisions. Browse the slides and follow the linked reading to examine how project knowledge can contribute to future competitive advantage.