Why I wrote a book: MBSE for AI, not AI for MBSE
Every engineering company is asking which AI tool to bolt onto its modelling stack. I think that is the wrong question, and I wrote seventeen chapters on why. What is in the book, what is finished, and who it is for.
The question I am asked most often by engineering leaders goes like this: which AI tool should we buy for our MBSE stack?
It sounds sensible. I think it is the wrong question. Not because the tools are immature, and not because the models are not capable. It is wrong because it assumes AI will assist a human author who keeps using tools built for human authorship. The model-based systems engineering tools most companies run were designed between roughly 2005 and 2012, for people clicking through diagrams. Adding an assistant to them does not change what they were built for.
The better question is the reverse: what does model-based engineering look like when AI is a primary author of the model, and humans review, decide and sign? That is the subject of my book, Sylang: MBSE for AI. Engineering models that AI writes and humans sign.
Why the title says MBSE for AI
The title reads the other way round from what the market is selling, on purpose.
| AI for MBSE | MBSE for AI | |
|---|---|---|
| Who serves whom | AI serves the modelling tools you already have | The modelling language is redesigned so AI can be its primary author |
| What changes | A language model is bolted onto the existing tool. Data model, formats and validation stay as they were | Plain text, strict types, a deterministic validator and a skill file for every file type |
| Who signs | Unclear: the assistant drafts inside a tool built for human authors | Engineers review, decide and sign |
| What you get | Faster first drafts, then schema drift, invented trace links and a painful audit | Speed that still holds up in front of the auditor |
The vendors are building AI for MBSE. The book is about MBSE for AI: changing the engineering model itself so machine authors can work in it safely, and people stay accountable for what ships.
What is in it
| Part | Chapters |
|---|---|
| For leaders | 1. Why MBSE-for-AI matters now |
| Foundations | 2. Model-based systems engineering, from zero · 3. The Sylang language, in one sitting · 4. Sylang Studio: your first project · 5. System project management with Sylang |
| The engineering spine | 6. Product line engineering · 7. Functions · 8. Architecture · 9. Requirements written for AI to consume · 10. Tests · 11. The engineering spine end to end |
| Safety | 12. The ISO 26262 safety chain · 13. FMEA · 14. FTA · 15. Traceability |
| Outputs and the argument | 16. Specifications and dashboards that update themselves · 17. MBSE for AI, restated |
Chapter 1 is written so a CTO or an audit committee member can read it alone and defend the argument. The later chapters are for engineers, AI builders and auditors, and each reader path is marked.
By the numbers
| Measure | Value |
|---|---|
| Chapters | 17 |
| Sections | 78 |
| Draft prose | about 80,000 words |
| Companion projects | 4: automotive electric parking brake, aerospace autopilot, medical infusion pump, industrial cobot |
| Companion model files | 28, about 2,900 lines |
| Standards covered | ISO 26262, DO-178C, IEC 62304, IEC 61508 |
| Format | 9 × 7 inch print, with an EPUB edition |
The electric parking brake is the anchor. It runs through the whole book, from the first hazard to the last test, so every concept is shown on one real system instead of toy examples. The other three projects show the same discipline in aerospace, medical and industrial settings.
Where it stands
The full draft of all seventeen chapters is done. What is left is the work that makes a book a book: completing the three secondary companion projects, front and back matter, a glossary and index, and final typesetting. I will share publishing details here when they are fixed.
Why a book, and why now
Because the decision most engineering organisations are about to make, which tools to buy and how to bring AI into the safety lifecycle, will shape their engineering for a decade. I would rather that decision be made with the full argument on the table than from a vendor demo. If you lead engineering, quality or AI in a safety-critical company and want to read an early chapter, write to me.