Why Casiana

Every mainstream learning management system stores one grade per assignment per student. That is a storage decision inherited from paper gradebooks, and it is the single biggest reason project-based schools find their software fighting their pedagogy.

The scalar cell is the problem

In a conventional gradebook the scoring record is a number. Rubrics, where they exist, are an input surface that sums criteria into that number — the rubric is the interface, the number is the record. You can tell because of what survives: export to CSV and the rubric is gone; look at a progress chart and it is a line of assignment scores.

That means the question a project-based school most wants to ask — how is this student doing at constructing explanations, across every class they take? — cannot be expressed, because no column anywhere means “constructing explanations”.

Teachers work around it by splitting one project into three assignments so the gradebook can hold three numbers. That fragments an integrated task into administrative pieces and teaches students that the science and the writing are separate things submitted to separate boxes.

What Casiana does instead

Design the assessment with the project.
A teacher tags an activity with the outcomes it assesses and the rubric to use for each. Students see the criteria before they start.
One deliverable.
The student submits one piece of work. No splitting, no duplicate submissions, no synthetic sub-assignments.
Score each outcome independently.
Each outcome gets its own score, its own rubric-dimension breakdown, and its own feedback. “Your reasoning does not yet connect your evidence to your claim” attaches to the thinking score, not to a general comment box.
Students self-assess against the same rubric.
Self-assessment is stored as a first-class score with the same shape, so a student’s rating sits next to the teacher’s, dimension by dimension.
Trends, not averages.
Per-outcome trend lines within a course, and a cross-course roll-up so a student’s agency or writing can be seen across all six of their classes at once.

The five outcomes we ship with

Defaults, not doctrine. Rename, reweight, remove, or replace any of them.

Knowledge & Thinking: Discipline-specific. Claim–evidence–reasoning in science, research and argumentation in English language arts, argumentation and explanation in social studies, problem-solving in mathematics.
Written Communication: Development of ideas, organization, and control of language conventions — scored separately, because “your writing needs work” is not feedback.
Oral Communication: Preparation, delivery, and interactive communication — including how a student handles a hard question from a panel.
Collaboration: Contribution, responsiveness to teammates, and follow-through, assessed while group work is happening rather than guessed at afterwards.
Agency: Goal-setting, persistence, help-seeking, and honest self-reflection. The outcome most often called unmeasurable, and the one students engage with hardest once they can see their own trend.

Questions we get asked

Do we have to leave Google Classroom?: No. Casiana is designed to be additive. Keep whatever handles day-to-day assignment distribution and use Casiana for the projects where multi-outcome assessment matters. A rip-and-replace migration is a bad trade for a school that only needs a better gradebook.
Can we use our own graduate profile instead of your five outcomes?: Yes. Outcomes, rubrics, dimensions, levels, and weights are all per-organization data, not hard-coded. The defaults exist so you can grade a real project in your first week rather than after a curriculum mapping project.
Does the AI assign grades?: No. AI drafts formative feedback for students before submission and can pre-fill a rubric for a teacher to correct. Every suggestion is labeled as AI-assisted, and a human teacher is the assessor of record. Submissions are stripped of identifying information before any model sees them, and AI interactions involving student data are written to the audit log.
What about student privacy?: Row-level security in the database is the security model, not an application-layer check — one school can never read another school’s rows. Access to student records is audit-logged with actor, action, target, timestamp, and origin. Data export and deletion are per-school and per-student capabilities, because FERPA requires both.
How big does a school need to be?: Small is fine. A single microschool of forty students works. So does a network of campuses that want to compare growth on the same outcomes without forcing every site to grade identically.

Join the waitlist

Pilot schools get direct access to the founder and a real say in what ships next.