An EM foundation model, with the solver as final confirmation.

Full-wave simulation is too slow to search with. We train a model on physics we generate ourselves, explore on it, and send only the survivors through to the solver.

Bring us a design Your geometry stays yours. On-prem or air-gapped.

What the model is

Not a chat layer over a solver. A learned model of the fields, built so that what physics guarantees is guaranteed in the model too.

FULL-WAVE LABELS

Trained on physics we generate

Our own label factory, drawn to cover the design space — not scraped from whatever got published.

E H (x, y, z, f)

Point-query, not fixed output

Ask for a field at a coordinate or a port-to-port response. One model, whatever resolution the question needs.

i j S(i, j) = S(j, i)

Symmetry built in

Mirror a layout and the fields follow. Reciprocity holds exactly — structural, not learned and hoped for.

You stop searching with the solver and start checking with it.

A full-wave solve is minutes to hours, so exploration gets rationed to a handful of candidates picked by intuition. Scoring thousands changes which designs you consider at all — and nothing ships on a prediction alone.

sweep 0912 · 2.4 GHz patchverdict.json

00:00 → feasibility reachable

00:01 → model_sweep 4,096 candidates

00:09 laminate A: band-limited — dropped

00:10 → promote top 6 → full-wave

00:52 → confirm convergence gates

00:58 every spec line met · pass

Design partners open

Four design spaces

Each loop takes a spec you already write down and returns a verdict saying which lines are met, on what evidence.

INSET-FED PATCH BAND TARGET MATCH vs FREQUENCY

Antenna

Planar single-element design with the laminate free to change — often it is the substrate, not the layout, that decides whether a spec is reachable.

You bring
Band, match, gain, efficiency, envelope, and the laminates you can actually buy.
The loop does
Screens feasibility, shortlists families, explores in bulk on the model, promotes survivors to full-wave FDTD with an independent FEM audit.
You get
Geometry, feed, and a verdict naming every spec line and the fidelity behind it. Impossible specs are called in a second.
DECOUPLED RAIL BEFORE AFTER TARGET IMPEDANCE vs FREQUENCY

Power delivery

Board-level exploration on the rail that matters, closed against a target impedance rather than a rule of thumb about decap counts.

You bring
A board and a rail: target impedance, the range it must hold over, and the parts you will place.
The loop does
Explores decap choice and placement, stackup and planes against the target, then re-derives the answer through the real toolchain.
You get
A changed design and one verdict for the whole chain, with the impedance result carrying the model behind it.
PAIR THROUGH A VIA MASK EYE AT THE RECEIVER

Signal integrity

High-speed channels end to end — the stackup, the via transitions and the routing that decide whether the eye is still open at the receiver.

You bring
A channel: data rate, the stackup you can build, and the connectors and package you are stuck with.
The loop does
Explores layer assignment, geometry and via structure against the channel budget, then confirms the survivors full-wave.
You get
A routed channel and a verdict on loss, crosstalk and eye margin, with the discontinuity costing you most called out.
COUPLED RESONATORS PASSBAND TARGET RESPONSE vs FREQUENCY

RF front ends

Distributed filters, couplers and matching networks synthesised against a response, rather than tuned by hand from a textbook starting point.

You bring
A response: passband, rejection, insertion and return loss, and the board area you can spend on it.
The loop does
Picks a topology, seeds it analytically, explores coupling and lengths in bulk on the model, then confirms full-wave.
You get
A layout and a verdict on every line of the response, including where laminate tolerance eats the margin.

Bring us one real design.

Thirty minutes: the spec you are trying to close, and the constraint that keeps biting. We run it and show you what comes back, including where it falls short.

Send us the spec