Skip to content
LEXUNCreate your KeyRun a decision free

What decision are you weighing up?

Describe it in your own words. The assistant separates what you actually know from what you are assuming, tells you which of those two matters most, and the released model runs it five thousand ways so you see a range rather than a single confident number. What you type here is sent to Anthropic to be read; what it receives is one sentence in the privacy notice.

FreeNo accountNo cardLIVE assistant connected on this deployment

ConversationStart from a question

Each one opens Quick decision with the question filled in. You type only the figures it asks for, on this device.

Optional AI: when you press Ask, the words you type here — figures included — go to Anthropic to be read. Your saved decisions and the engine's inputs never do. How it is handled.

Decision record

Decision
not yet stated
Model
chosen once the decision is clear
Facts
none confirmed
Assumptions
none declared
Still needed
—
Result
the engine has not run

Every figure here is one you typed. Nothing is filled in for you, and nothing is sent anywhere until you press Ask — then your words go to Anthropic to be read. A figure asked for afterwards stays in this browser. After a run, the Copilot’s Explain, Test my assumptions and Board summary send it the result’s summary — the band, its percentages and the names of your inputs, never the figures. Nothing else leaves this browser.

What this can and cannot answer. Three models are released, and each answers a money question with a stated boundary:
  • Business runway — whether cash lasts. It does not forecast your sales.
  • Career change — whether your buffer survives the gap. It does not judge the job.
  • Major purchase — whether the commitment fits. It does not value what you are buying.
Anything outside those, it should refuse rather than force. If it ever gives you a number without a receipt you can reproduce, that is a defect — check any result here.