The AI-native engineering company for the physical world.
Engineering studies for regulated manufacturers, delivered on a model of the site built from what is actually installed.
We do the engineering work a plant buys across its life, starting at the front end.
Feasibility, pre-tender baselines, handover verification, change-control assessments, turnaround readiness and due diligence. Each study starts from a model of the site that our engineers build from the drawings, scans, photographs and records the site already holds, reconciled against what they find on the walk.
Where the drawing and the plant disagree, the study says so before the tender goes out. Design on the verified plant is the next rung. Today the deliverables are studies, registers and verdicts, and the site's engineer of record signs them.
Where the first purchase comes from
Two doors, depending on what the site has in front of it.
A new line or building in its front end
The first purchase is SyntaGraph attached to the project: the evidenced site model, the factory model and the discrepancy register the EPCM bids against. The project manager buys it from the project's existing-conditions budget, the line that already pays for surveys and site data. Once we are an onboarded vendor on the project, the engineering studies follow as change orders.
A customer can start with the model alone. We prefer a study alongside it, because the verdict is what closes the record, but we do not make it a condition.
SyntaGraph on a capex projectNo big project, one awkward question
Can this equipment fit here? Where there is no big project, a verdict on one placement, at a fixed scope and a fixed price, is how the work starts. Our engineer walks the bay, checks footprint, clearance, access and utilities against what is installed, and returns the answer together with the discrepancies found on the way.
When the question is bigger than one bay, it becomes a feasibility study on the same site model.
The fit questionThree fronts on one site model.
VN Engineering
Front end, pre-tender, handover, change control, turnaround and due diligence. The studies a project engineer commissions, delivered on the plant as built.
Engineering
VN Supply Chain
Capacity, downtime and transfer questions answered on the same site model, from the planner's own exports, validated fact by fact.
Supply Chain
VN Applied Labs
Environments, evaluations and verifiers for AI builders, built from synthetic plants. Customer sites stay the private exam.
Applied LabsThe plant you can ask what happens if.
Every study starts from the site as evidenced. The same site can be asked what a change does to the air, the heat and the utilities around it: whether the new line creates a hotspot in the bay, whether the extract holds the pressure cascade, whether the utility headroom is really there.
Each boundary condition points to its evidence, and each result is labelled screened, verified, measured or signed. What your product does inside the machine stays with your process engineer, and the study says so.
This capability is being built into SyntaGraph. Until it exists, a proposal states which questions are answered by declared hand calculation and which are Not Demonstrated.
One bay, drawn by code to show the method. The session is scripted and shows no real result.
The record outlives the project.
A verdict keeps its evidence, its conditions and the approval that followed. The discrepancy register from a handover check is the one the site would otherwise carry for years. Every later question on that site starts from the record, which is why the second study costs less than the first.
- ObservedPhysical state
- AttachedEvidence
- FoundConstraints
- SignedThe engineer's decision
- DoneAction
- MeasuredOutcome
When the evidence cannot carry a conclusion, the verdict says Not Demonstrated and names the evidence that would close it. We price and deliver that answer like any other.
Tell us about the project.
Written for project engineers and project managers with a line, a building or a placement in front of them, for supply chain and site leads with a capacity question, and for AI teams that need a harder physical test.