Physics-aware AI agents, with an engineer in the loop.
LumiMerse is Velon’s physics-aware agent engine for energy and engineering teams. It turns manual and semi-manual engineering work into robust, end-to-end, fully auditable production workflows — connecting simulation platforms like PSCAD and PowerFactory, Python, APIs and open network data when the job needs them — with an engineer review gate before anything leaves the system.
A physics-aware agent engine for auditable engineering tasks.
Across these industries, critical engineering work still runs manually or semi-manually — data pulled by hand, solvers driven one case at a time, results assembled into reports nobody can fully trace. LumiMerse changes that. It builds robust, end-to-end production workflows that are auditable by construction — orchestrating whatever the job needs: PSCAD/EMTDC, PowerFactory and Python when simulation is at the core, or APIs and open network data when it isn’t. The same study runs the same way every time, and every result carries a traceable record of how it was produced.
It is deliberately not autonomous. LumiMerse prepares, runs and checks; an engineer reviews and signs off. Nothing reaches a client or a regulator without passing a review gate. Every run is logged, every input is versioned, and every output traces back to the requirement it addresses — which is what makes the result defensible when a reviewer asks how it was produced.
Six stages. Click any of them.
One review-gated pipeline underlies every workflow — what the engine does, and what gets recorded, at each stage.
Brief & scope
MandatoryThe engine does
Recorded for audit
Where teams put it to work.
Illustrative workflows. Select one to see the shape of the work and the output it produces.
LumiMerse, applied.
Two demonstrators built on Velon's own work — the engine doing real engineering, not describing it.
Smart Agent for Grid Reinforcement
A LumiMerse workflow that augments the tools planners already trust — DIgSILENT PowerFactory and Python — and automates the cognitive labour around network reinforcement: deciding what to study, interpreting results, screening compliance and drafting the report. It does not replace the solver. It replaces the week of manual work around it.
An orchestrated run
A coordinator plans the study, delegates to specialist agents, and routes every result through an engineer review gate. Press play to watch a run, or tap any agent to see what it does.
Consumes
Produces
Study setup & execution
Builds the scenarios and runs the batch — no manual model wrangling.
Results triage
Finds, ranks and explains the violations that actually matter.
Compliance & reporting
Screens against your criteria and drafts the report.
Engineer-in-the-loop
Every step is reviewed and approved by your engineers.
Just one example of what LumiMerse can do. Tell us the manual process that slows your team down — we build the workflow that removes it.
Stability Potential
An evidence base, a working model and a procurement-ready scorecard for stability services delivered from below 132 kV — turning the invisible distributed fleet into a qualified source of system strength and inertia for the transmission operator.
The largest unmodelled stability resource sits below 132 kV.
System strength and inertia are procured almost entirely from synchronous plant on the transmission network. Below 132 kV, tens of gigawatts of distributed energy resources are effectively invisible to the stability toolkit — capable of providing the service, but never qualified to.
The distributed fleet
Each technology below 132 kV can provide some stability services — but not all, and not equally. Tap an asset on the left to see how much of each service one unit can offer; tap the GSP hub to see the fleet aggregated against the operator’s target.
Capability evidence
Which distributed assets can provide system strength, inertia and damping — per technology.
Validated model
Performance proven at the connection point through RMS/EMT study.
Procurement scorecard
Qualified potential turned into a service the operator can actually buy.
A new resource
Tens of GW below 132 kV, made visible and dispatchable.
A bespoke study and toolset built for a real operator question. Bring us yours — we'll build the analysis and the tool around it.
See a workflow on your problem.
Tell us the workflow you want to make repeatable — a compliance study, a validation task, a study pack — and we'll show you how LumiMerse would structure and govern it.
Governed, auditable workflow automation for power-system engineering teams. We'll be honest about what is production-ready today and what is on the roadmap.
What agentic AI means here.
What does “agentic AI” mean in the context of power systems?
How do physics-aware agents differ from generic LLM prompts?
How do these workflows stay auditable and regulator-friendly?
Brief us on a project.
Send a one-paragraph scope — the system, the codes that apply, and the question you need answered. We typically reply within two working days.
Related: Consultancy · Velon Suites · Network reduction
Skip the form — email the team directly.