What AI4MBSE is
AI4MBSE is Ark flagship conversational SysML / MBSE modeling assistant. Grounded by the engineering knowledge base and MBSE RAG, it lets engineers describe intent in natural language, advance one diagram at a time, review checkable candidates, and write back to hosts such as MagicDraw, Cameo, Capella, or SysON only after human confirmation — human-in-the-loop write-back, not one-shot whole-system generation.
Problem it targets
MBSE programs stall when knowledge lives in documents, when only a few experts can navigate SysML tools, and when generic LLM chat hallucinates structure that fails review. AI4MBSE sits beside the modeling environment you already use, respects methodology rhythm, and retrieves project conventions so suggestions sound like your program — because they are constrained by your KB slices.
Core capabilities
| Capability | Engineering meaning |
|---|---|
| Conversational modeling | State the sprint goal; assistant advances the active view |
| MBSE RAG | Retrieve standards, cases, and rules by method + diagram type |
| Checkable candidates | Inspectable structure; explicit unknowns |
| Human confirm → write-back | People own baselines; AI accelerates drafting |
| Kernel + adapters | Swap hosts without rewriting the assistant core |
Difference from chat-only LLMs
Chat models optimize fluent language. AI4MBSE optimizes modeling rhythm: which diagram is in scope, which elements may be created, which retrieval fragments apply, and which write-back API calls are allowed after confirmation. It will refuse to silently invent trace edges or requirement IDs when the KB does not support them — engineering honesty over marketing fluency.
Architecture: kernel and hosts
The kernel owns dialog orchestration, intent detection, retrieval against the engineering knowledge base, candidate SysML structures, and write-back contracts. Host adapters connect to desktop tools via plugins, to open-source stacks such as Capella and SysON, or to customer gateways and SysML v2 API endpoints. Product direction embraces SysML v2 exchange while hardening adapters teams use today.
Relationship to the knowledge base
The KB is not a generic document pile. Curators slice content by methodology (MagicGrid, OOSEM, Harmony-SE), diagram type, and project convention. At task time, MBSE RAG pulls the smallest useful context — enough to constrain candidates, not enough to drown the session. The KB is also planned as an independent product serving simulation and other apps; see the Knowledge base page.
Typical session workflow
- Engineer selects host project and states diagram goal.
- Assistant retrieves KB fragments and prior model context.
- Candidate structure renders for review with diffs where possible.
- After explicit confirmation, write-back executes in the host.
- Session log retains prompts, citations, and actions for audit.
Methodology awareness
Teams follow different MBSE paths. AI4MBSE does not force a single notation gimmick; it loads methodology profiles so MagicGrid domains, OOSEM operational views, or Harmony-SE use cases shape retrieval and pacing. Read more on the Methodology page and explore industry reference scenarios.
Tooling stance
MagicDraw and Cameo remain common in defense and rail; Capella and SysON matter for open architecture work; SysML v2 text opens CI-friendly routes. Ark does not ask you to rip out tooling — adapters meet you where models live.
AI4MBSE online trial
The trial ships productized capabilities from real MBSE work — not a hollow demo. Roadmap: practice → plan → refactor → open trial.
2026 polishing window
- 01Practice — Prove dialog modeling, KB grounding, write-back in real work.
- 02Plan — Product boundaries and repeatable delivery.
- 03Refactor — Harden kernel, contracts, adapters.
- 04Open trial — Public entry when ready — updates on this page.
Updates via FAQ; the public entry will appear on this page when ready.
Reference scenarios
-
Platform screen doors (PSD)
Industry reference scenario — read methodology context and modeling focus.
-
Intrinsically safe mine personnel positioning base station
Industry reference scenario — read methodology context and modeling focus.
-
Robot vacuum obstacle avoidance
Industry reference scenario — read methodology context and modeling focus.
-
Onshore wind farm power & safety interlocks
Industry reference scenario — read methodology context and modeling focus.
-
Portable infusion pump control unit
Industry reference scenario — read methodology context and modeling focus.
-
Warehouse AGV avoidance & dispatch
Industry reference scenario — read methodology context and modeling focus.
Author & standards context
Yuan Liangding builds AI4MBSE in Chengdu, drawing on industrial platform delivery, knowledge engineering, and hands-on AI-assisted MBSE experiments. Public discourse from INCOSE on model-based competencies, OMG on SysML v2, and open communities around Capella and SysON inform the architecture — without implying endorsement.
Business value of conversational MBSE
Programs pay for time-to-clarity: how fast a multidisciplinary team can turn verbal agreements into reviewable SysML artifacts that survive configuration management. AI4MBSE shortens blank-canvas time, keeps terminology aligned with the curated knowledge base, and leaves authority with reviewers. That combination — MBSE RAG plus one-diagram pacing plus human-in-the-loop write-back — is the product thesis, not a chat novelty.