MBSE · AI-native · Human-in-the-loop · Safety-critical
The engineering intelligence layer
for safety-critical programs
VectorMBE is an AI-native MBSE platform that connects your engineering toolchain into a governed knowledge graph. AI agents draft requirements, synthesize safety assessments, trace change impact, and compare design alternatives. Engineers review and approve every output.
The problem
Programs fail at the seams between tools
Requirements live in DOORS. Architecture lives in Cameo. Test evidence lives in a spreadsheet. When a requirement changes, the downstream impact has to be traced manually across all three. By the time anyone spots the gap, months of design work are already committed.
of program cost growth attributed to requirements instability (INCOSE 2023)
of defects traceable to poor requirements definition in safety-critical systems
disconnected tools the average aerospace program uses across the engineering lifecycle
What VectorMBE does
Six capabilities, one governed graph
Requirements generation
AI agents extract and draft formal requirements from source documents, SysML models, and signal databases. Every requirement traces to its source. Engineers review and approve before anything enters the graph.
Safety assessment synthesis
Generate structured safety assessments from the graph. Failure modes, hazards, and mitigations link to the requirements and components they cover. Constraint gates prevent safety-critical items from being bypassed.
Architecture reasoning
Query the graph to understand how components relate, which functions satisfy which requirements, and where interface gaps exist. Natural language queries resolve to graph traversals across the full system model.
Change impact analysis
When a requirement or component changes, graph traversal identifies everything downstream: interfaces, tests, verification items, and affected safety assessments. Nothing falls through the cracks because the graph knows the dependencies.
Verification tracking
Test results, inspection records, and analysis artifacts attach to the requirements they verify. The graph surfaces open verification gaps and shows closure status across the program without manual status collection.
Constraint-driven trade analysis
MCP-connected AI agents compare design alternatives against constraints and objectives. Each run logs results with full provenance. Agents cannot override constraint gates, so the governed graph stays consistent throughout.
Industries
Built for programs where failure is not an option
Aerospace
DO-178C, ARP4754A, and AS9100 programs. Requirements traceability from system to software, with verification evidence linked at every level.
Defense
MIL-STD and DoD digital engineering programs. Formal ontology aligned to DoD SERC handbook standards. Graph-based system of record for program offices.
Automotive
ISO 26262 and ASPICE programs. Signal catalogs, DBC/ARXML ingest, and AUTOSAR interface traceability. Requirements exportable directly to DOORS and Jama.
Infrastructure
Critical infrastructure and industrial control programs. OWL-based system models with formal constraint enforcement for long-lifecycle asset management.
Toolchain integration
Works with the tools you already use
VectorMBE connects to your existing toolchain via the Model Context Protocol and standard interchange formats. It does not replace your tools. It governs the space between them.
Requirements
DOORS, Jama, PTC Integrity, ReqIF export
Modeling
Cameo, MagicDraw, SysML v1/v2, XMI ingest
Signals
DBC, ARXML, CSV signal catalogs
AI assistants
Claude Desktop, Cursor via MCP
Ontology
OWL 2, RDF/Turtle, BFO, SOSA alignment
Documents
PDF, Word, specs: ingest and extract
Get started
Install VectorMBE
The repository is currently private while we finalize the public release. GitHub Sponsors get immediate repo access and can build locally today. Docker is the fastest path.
Docker (quickest)
Pulls the pre-built image from Docker Hub. No Rust toolchain needed. Works on macOS, Windows (WSL2), and Linux.
git clone https://github.com/radsilent/vectormbe-deploy.git vectormbe
cd vectormbe
cp .env.example .env
# Edit .env and set VECTORMBE_LICENSE_KEY
docker compose up -dOpen http://localhost:8080. See the full deploy guide for GPU options and production setup.
Connect via MCP
Register the MCP server in Claude Desktop or Cursor. Run the script for your platform, then restart the app.
{
"mcpServers": {
"vectormbe-runtime": {
"type": "stdio",
"command": "vectormbe-mcp",
"args": [],
"env": { "VECTORMBE_LOG_LEVEL": "info" }
}
}
}See the MCP setup docs for platform-specific scripts.
Import formats
Ready to close the gaps in your engineering program?
We work with aerospace, defense, automotive, and infrastructure teams. Talk to us about your program.