Find reusable engineering designs with an ontology connecting requirements and evidence.

When your engineers search for an existing design that meets a new requirement, finding a familiar drawing still leaves them tracing its revision and supporting evidence. Connecting requirements, components and design records through an ontology can help them inspect potential reuse candidates together, with the evidence needed to decide what deserves further evaluation.

An ontology is a shared model of business concepts and their relationships. For engineering retrieval, those concepts include components, requirements and engineering models—the kinds of information connected in the Siemens XSearch implementation account.

The business opportunity is less reconstruction before engineering judgment begins. Instead of asking someone to assemble the context separately for every search, a connected retrieval workflow can present a candidate alongside the records supporting its relevance. Functional AI Solutions would build that workflow around a foundational business ontology and connected workflows, repository connectors and links back to source evidence. Your engineers retain responsibility for deciding whether a design is suitable.

Find useful context, not just matching documents

Engineering retrieval becomes more useful when a result explains how a component, requirement and design record relate. A matching filename gives an engineer somewhere to start; a candidate linked to the relevant drawing revision and test report gives them a more concrete basis for assessment.

The proposed ontology would distinguish the component from its design revisions and record which evidence supports a candidate relationship. The retrieval interface would present those records together and flag unresolved requirements or conflicting revisions. That is the mechanism behind the intended benefit: less time reconstructing context, with more of the engineer’s attention available for evaluating it.

Your existing repositories remain the sources of engineering records. FAIS would connect the selected repositories and model the relationships needed for retrieval, rather than make replacement of those systems the pilot’s objective. Connectors and source identifiers still need to be implemented so the interface can retrieve the correct records.

What can you learn from Siemens XSearch?

Siemens XSearch illustrates how a shared engineering model can connect information across applications, CAD data and documents. In its account published on September 2, 2026, implementer Semantic Arts reports that the initial ontology extended its gist foundation and contained approximately 50 core concepts. The modeled scope included equipment, components and systems; functions, specifications and requirements; engineering models; and bills of materials. Read the implementation account.

Semantic Arts also reports that the effort developed beyond its original search proof of concept into a broader semantic architecture for engineering and manufacturing, providing a foundation for digital-twin work. Source

The useful lesson for your business is to model relationships that serve the first search workflow and can also support subsequent applications. Define your pilot around an engineering question, rather than adopting the concept count of another engagement.

Bring the candidate and its evidence together

Consider an illustrative Taiwan manufacturer assessing whether an existing pump assembly could support a new operating requirement. Before connected retrieval, an engineer searches past projects, opens drawings, checks component records and locates a test report. They then reconstruct which component belongs to which assembly, which design revision the report covers and which requirement the evidence addresses.

After the proposed workflow is implemented, the engineer would receive a candidate revision with links to its drawing, relevant report and the requirement those records may support. Missing evidence and revision conflicts would remain visible rather than being treated as a match.

A candidate is a starting point for engineering judgment, not design approval. The engineer would still assess operating conditions, unresolved requirements and revision compatibility before reuse proceeds.

FAIS’s role would be to model those distinctions and connect the retrieval workflow to them. Keeping the relationship definitions separate from a single search screen also gives later workflows or agents a model they can reuse, with suitable connectors and implementation. The first investment can therefore establish useful structure beyond the initial interface without requiring a company-wide rollout.

Fund expansion on engineer-confirmed usefulness

An engineering-reuse pilot should test whether connected retrieval makes useful prior work easier to find. Agree the evaluation before building:

1. Select representative historical engineering questions within one product family.
2. Record baseline search effort using the current repositories and process.
3. Repeat retrieval with connected records and document the test conditions.
4. Have engineers assess candidate relevance and the adequacy of supporting evidence.
5. Record missing evidence, unresolved requirements and revision incompatibilities alongside useful results.

Search effort and engineer-confirmed relevance belong in the same assessment. Set an acceptance threshold before testing: a faster search is useful only when its results help engineers evaluate the work.

Your practical starting point this quarter is one product family, a responsible engineer and a small set of representative search questions. Identify the repositories and relationships those questions depend on. If engineers repeatedly reconstruct the same connections, that gives you a concrete reason to discuss an ontology-backed retrieval workflow with FAIS—and a bounded way to evaluate whether it merits expansion.