Marko quietly learns a few things about how you like to teach so that future marking and conversations feel more like you. This article explains what Marko remembers, how it learns, and how that memory is used.
For how to view or edit what Marko knows, see Editing or clearing Marko's memory.
Marko's memory is split into three named slots:
How strict you want the AI to be when marking. Marko picks up on patterns — if you regularly override marks downward (or upward), Marko notices and may suggest saving that preference.
The default tone for AI-generated feedback. Warm and encouraging, brief and businesslike, or somewhere in between — Marko remembers your preference and applies it on future submissions.
A free-form text area for anything else that's useful. "I teach IB Biology HL, Year 12." "Always award marks for showing workings." "Don't penalise British spellings in US students." Anything that helps Marko and the marking engine understand how you prefer to mark.
There are three ways something ends up in Marko's memory:
Marko's memory is shared with the marking engine. Strictness and feedback tone in particular affect automatic marking on new submissions — not just Marko's chat responses. That means a small change in your preferences ripples through your whole experience.
Marko's memory is intentionally narrow. We don't remember: