Education
A course instructor who wants every student to get feedback that references what that student actually submitted before.
[ 01 ]The problem
Feedback at scale collapses into rubric boilerplate. The instructor knows student #43 made this same mistake in assignment two — but holding ninety trajectories in your head is impossible, so feedback resets every week and progress goes unseen.
[ 02 ]How Mynd handles it
Each learner gets a trajectory
Y0's context graph keeps a per-student record: past submissions, past feedback, recurring error patterns — the longitudinal view no one has time to maintain.
Feedback is planned against the rubric and the history
Each submission is assessed against the rubric and that student's trajectory, so the draft says 'same off-by-one pattern as assignment 2' instead of generic advice.
The instructor reviews, the runtime drafts
Drafts arrive with the evidence attached — the rubric line, the prior submission — and the instructor edits at the level of judgment, not typing.
[ 03 ][ example run ]
plan ✦ review 87 submissions — assignment 4 context ✓ rubric v3 — loaded context ✓ trajectories × 87 — loaded execute → draft feedback, history-aware execute → flag 6 students: regression pattern result ✓ 87 drafts · 6 interventions surfaced · evidence attached
[ 04 ]more feedback depth per teaching hour