Project brief
Instructors and academic units need help preparing class, giving feedback, and noticing struggling students — without an autopilot that changes grades or contacts students. Atlas is that intelligence layer: AI suggests; humans decide.
Instructors bring syllabus, PDFs, rubrics, and assignments. They get prep drafts and assessment feedback workflows immediately. Connecting Canvas (or another LMS) is an upgrade for live rosters, submissions, and later write-back — not a gate.
Private workspaces (Supabase Auth + RLS), course materials upload/preview, Prepare (lesson drafts grounded in materials), and Assessment Intelligence (score/feedback drafts with evidence and human review states). A separate sample multi-role demo remains for department walkthroughs.
No published conversion metrics, grading accuracy benchmarks, or institutional deployments beyond the live pilot. Claims stay at capability and architecture level until measured results are ready to share.
Built by Carlos López González. Source repository is private; public entry point is the live pilot.