Three tiers. One conversation to start.

No. Structurally. Two horizon guards — 25.2 hours on the data layer, 25.0 on the solver — are imported into every acquirer, every builder and the streaming bridge. A future-dated record is refused at ingestion; a query past the cutoff returns INFEASIBLE_HORIZON. There is no flag to turn this off. This is load-bearing, not small print: it's why the tool is auditable, why it has no market-abuse surface, and why you can put its output in front of a regulator.

Yes — that's the Core licence working as intended. JSON topology maps and IEEE/MATPOWER case import are in every tier. Your network gets the same solver, the same exact reactive limits and the same math-ladder exports as the national grid does.

The asset data is real — buses, demand, generation, transformers, from actual DNO and NESO publications. The transmission spine's wiring is a physically-reconstructed topology over the real GSP sites, because NGET doesn't publish its connectivity. It re-validates to machine epsilon either way. The full statement is on the GDA page's disclosure card, where it belongs — not buried in a footnote.

Byte-for-byte. The solve export from the same job is identical every run — a test asserts it. Dense and sparse backends produce bit-identical voltages; GPU results are bit-for-bit the CPU's. If your reviewer re-runs your study, they get your bytes.

30 days, full-featured, at every tier — including the GDA module. Evaluation exports are watermarked; everything else is the real thing. If the maths doesn't convince you in a month, no pitch was going to.

System-operations research groups. University power-systems departments. Consultancies who need defensible historical grid states rather than hand-waved ones. Regulators' technical teams who want to check a claim against the record. Anyone whose question begins "what was the grid actually doing when…" — and nobody whose question begins "what will".