The engine, live

Answer five. Watch it read why.

Pick an answer, then say how sure you were. The engine reads the two together and sorts each miss into one of four kinds, building a learning record in your browser as you go. It is the same engine the API runs. Five answers is enough to show how the reasoning works — not enough to profile a learner, and the report says so.

Synthetic demonstration · nothing stored · no account

Some readings show n/a when five answers do not yet give enough evidence to estimate them. That is the engine declining to guess, not an error. With more answers the picture fills in.