ACADEMY
Neurosymbolic Systems Engineering
From ontological modeling to AI-assisted engineering, built on open standards with hands-on practice.
- Start anytime, no cohort to wait for
- Course slides and recorded videos
- 1-hour live office hour every week
- Hands-on exercises and full project, with feedback
- 6-month evaluation license of OML Code
WAYS TO TAKE IT
Choose How to Enroll
Asynchronous
8 weeksStart anytime
Enroll$3,500$2,500USD + applicable taxesIntroductory price for early enrollments- Talk to us
Live Online or Onsite
A private cohort for your team
Payments are processed by Paddle, our merchant of record. Your receipt and invoice come from Paddle, and sales tax is calculated at checkout. Your statutory 14-day refund right ends when the course materials are first shared with you, or 14 days after purchase, whichever comes first.TermsPrivacyRefund policy
WHY IT EXISTS
Your engineering knowledge is trapped in documents
A system's design lives across a requirements tool, a modeling tool, spreadsheets, scripts, slide decks and test reports. Nothing connects them, so the questions that matter most, which always span two of them, take days to answer and are answered with caveats.
- Which requirements does this design actually satisfy?A person reads both and builds a trace matrix by hand. It is correct the day it is delivered.
- If I change this interface, what breaks?Ask the people who would know. The answer is as good as their memory of what they last touched.
- Is this design self-consistent?Nothing checks it. The contradiction is found in integration, or by a reviewer who happens to recall the other half.
- Show me the current state of the design.Somebody spends a week assembling slides. By the review, part of it is already out of date.
WHAT A WEEK LOOKS LIKE
Eight weeks, and every one of them marked
You work at your own pace, on two models at once: the running example, and a system of your own choosing that you build a methodology for as the course goes.
- 1StudyWork through the week's slides and recorded video on your own schedule, shared from Google Drive.
- 2ApplyAssignments on the running example, a wildfire-response system of systems, where each concept is used on a system you already know.
- 3BuildApply the same step to a project you choose, developing your own methodology and case study a layer at a time.
- 4ReviewBoth deliverables are read by the instructor every week, with written feedback before the next one starts.
- 5DiscussA live office hour each week to take up whatever the feedback raised.
Delivered directly by Modelware. You will hear from us after you enroll. Slides and videos are shared from a Google Drive, feedback on assignment/project are communicated by email, and your questions are answered on Slack and in live office hours.
WHAT A NEUROSYMBOLIC METHODOLOGY IS MADE OF
The pieces you will build and run
- VocabulariesThe concepts of your domain and what they mean, in a logic rather than a glossary.
- ShapesWhat a model must contain to be complete and correct, checked on every change.
- ViewpointsTables, trees, graphs, matrices, charts, diagrams, and custom views that answer one stakeholder's question.
- AnalysesQueries and computation over the graph: coverage, impact, margins.
- AI agentsAI that respects and is empowered by your defined methodology rather than blindly or randomly uses files.
- Code generationExecutable artifacts from the model, including behaviour you can simulate.
- CI/CDEvery commit validated and reasoned over; every release regenerates its products.
- FederationModels from several repositories integrated incrementally by carefully configured dependencies.
- Dependency managementMethodology and model slices as versioned packages with resolved dependencies.
- Configuration managementModels are text: branch, review, merge, tag a baseline, reproduce an analysis.
THE NEUROSYMBOLIC LOOP
Generate, check, repair, repeat
A fluent AI assistant and a formal reasoner, each doing what the other cannot, with the methodology between them as the contract.
- 1InterpretAn AI assistant reads prose, documents and intent and identifies key words to ground.
- 2ContextualizeThe words are grounded in the model and the methodology, so they are unambiguous.
- 3ProposeIt drafts a change in the models and analysis notebooks based on the defined methodology.
- 4CheckA reasoner tests consistency, constraints and what follows.
- 5RepairPrecise errors go back as instructions, and the loop closes without you.
- 6ExplainThe assistant puts the reasoner's derivation into language you can read, so what you are approving is the argument and not just the answer.
Neural: AI assistantSymbolic: model and reasonerYou