PROSTEP Digital Thread Platform: Structured engineering knowledge for trustworthy AI agents

Space programs depend on information spread across many systems: requirements in ALM tools, SysML models, CAD and BOMs in PLM, manufacturing data in ERP and MES, plus test and qualification evidence. Each tool is the right choice for its domain. But none of them stores how its data relates to the data in the other systems.

The PROSTEP Digital Thread Platform closes this gap. It connects PLM, ALM, ERP, MES and MBSE tools through more than 30 production-ready connectors. Every synchronization stores typed, versioned relationships in a central Knowledge Base. The result is a Product Information Graph based on the prostep ivip Top Level Ontology and the CASCaRA model. Data stays in its source systems, and the graph links to the original objects. AI agents help build the graph by proposing links between requirements, components and tests. Engineers review and confirm each link before it becomes verified knowledge.

On top of this, the PROSTEP AI-Workbench lets AI agents work on verified, version-accurate configuration items via the Model Context Protocol (MCP). It follows the MADWYN principle (Make AI Do What You Need): graph traversals and rule checks run deterministically, and the language model only handles interpretation and wording. This makes results reproducible and auditable. The AI-Workbench is LLM-agnostic and can run with local models on-premises, in a private cloud or as SaaS. Teams increase agent autonomy step by step, from Advise to Assist to Automate.

What this enables:

• Cross-domain impact analyses in seconds instead of days

• Configuration states traceable from as-designed to as-built and as-maintained

• Compliance evidence produced by graph traversal instead of manual collection

• Natural-language questions answered with links back to the source systems

• Secure, logged data exchange with partners, including ITAR-relevant scenarios

We’re looking for space OEMs, suppliers and research partners who want to put engineering AI on a reliable data foundation, as pilot users or in joint R&D projects.