🗺️ Use cases

Where an evidence-backed model earns its keep

Three stories from industry and one from the Moon. Different sectors, same question underneath: what does this control system actually do — and can you prove it?

The industry stories are illustrative composites of the situations we build for, not customer references, and every sample answer comes from a synthetic system — we never show customer data.

1202
🌓 The fun one

Apollo 11, and the most famous error code ever flown

The lunar module's source code is public domain — so we put a DSKY on the page. Ask Daritas what a 1202 program alarm is, and check every citation against the lines that flew.

Try it →

And then the assessment is over. That's when it starts earning.

Every story above ends the same way: a model that stays. Four things that happen next, in every one of them.

Six months later, 03:00, an alarm nobody recognises

Ask. The answer comes from that system’s own assessed model with the evidence behind it — not a general AI guessing from a code snippet, and not a binder nobody can find.

Someone changes the logic

Re-ingest. You get a diff: which findings still hold, which were invalidated, and what needs re-review — scoped to what actually moved, not the whole plant again.

Your team already works in their own tools

Read the verified model over an authenticated API, or let an AI copilot query it over MCP — read-only, scoped to your organisation by your own key.

The engineer who knew it retires

What they knew was captured against the claims it explains, attributed and dated. The model is yours in an open format — it stays readable whether or not we are still in the picture.

Your sector isn't here?

If it runs on control logic — plant floor, building, machine, or something stranger — it fits. Tell us what you run.

Request beta access