Cogitan

Instruments

Built to do our own work.
Sold because they hold up.

Every one of these exists because we needed it for a question we were trying to answer. That ordering is why we can tell you where each of them stops.

The problem we solve

Simulation doesn't scale.

Full-fidelity simulation is accurate but slow. A 1,000-cell RSFQ block takes 5–10 minutes in SPICE; a full chip is an engineering day, per iteration. The constraint isn't compute — it's the simulation paradigm itself.

Speed
Sequential. Minutes per block, hours per chip.
Batch-parallel. Milliseconds per cell.
Confidence
One answer. No uncertainty. No ranking.
1σ bounds on every output. Low-confidence cells auto-flag.
Scale
Every candidate requires full simulation.
Screen 10,000 candidates. SPICE validates the shortlist.
Optimization
Not differentiable. Gradient-based optimization is impossible.
Fully differentiable. Backpropagate to any target metric.

How an engagement runs

01

Share your simulation data

Cell library and reference simulation outputs for your process node — MIT-LL SFQ5ee+, IPHT, SeeQC, AIST, SkyWater, or proprietary.

02

We train the surrogate

Calibrated to your junction parameters, cell geometry, and design rules. Training runs on your data.

03

Explore at ~10³×

Design-space exploration with per-cell confidence scoring. Low-sigma predictions pass; high-sigma cells are flagged.

04

Validate the shortlist

Your reference simulator validates the finalists. Every shipped design is ground-truth verified.

Access

Fluxus and Corpus are delivered as an API and CLI, trained per process node. Readout Co-Design is scoped and priced before any work starts. All three are calibrated to your process rather than sold as a general model.