Scaffold — Trust, Privacy, and Compliance

The instructor is never removed from the loop.

Scaffold is a tool that works under the instructor's authority, not a system that replaces their judgment. Every measure we take is designed to scale human judgment — never to replace it.

Scaffold groups similar student responses and presents representative samples for the instructor to grade blind — without seeing the AI's assessment first. Scaffold's own evaluation of each submission comes from multiple independent assessments designed to ensure fairness, and adapts to the instructor's corrections throughout the process. Where the instructor's judgment and Scaffold's diverge, the instructor's decision always takes precedence.

Submissions where the system is less confident — outlier answers or responses that don't fit neatly into a group — are surfaced in a dedicated review queue for closer human attention. And in every case, final approval cannot be bypassed. No grade is released to a student until the instructor has reviewed and approved it. Scaffold doesn't produce final grades — instructors do.

Your whole team, one standard.

Most courses are graded by teams of instructors and teaching assistants, each bringing their own interpretation of the rubric. Scaffold ensures that every grader on your team applies the same standard. TAs can be assigned to specific questions, so their judgment is focused and consistent. Role-based permissions ensure that each team member sees only what's relevant to their role — TAs access the assessments and questions they're assigned to, instructors maintain oversight of the full course, and students see only their own grades and feedback.

Our AI never sees who your students are.

When Scaffold's AI evaluates a student's work, it receives the answer text and the rubric — nothing else. Student names, email addresses, student IDs, and other identifying information from our database are architecturally excluded from every AI grading call. The AI assesses student work on its merits, without knowing whose work it is.

Student answers are sanitized before transmission and wrapped in protective delimiters to guard against prompt injection. The separation between student identity and student work isn't a policy — it's enforced by the system's architecture.

Every action on the platform has a paper trail.

Scaffold maintains a comprehensive audit trail that extends well beyond grading decisions. Every login, every file uploaded or removed, every grade override, every approval — each action is logged with a timestamp, the user who performed it, and a before-and-after record of what changed.

At the grading level, each score traces back to specific rubric criteria, a documented rationale, and a confidence assessment. When an instructor overrides a grade, the original score, the new score, and the reason for the change are all preserved. Platform states are captured so that subsequent changes — such as course files being added or removed after grading — don't leave gaps in the record.

Every grade can be defended with evidence, not just professional judgment.

Students have a structured path to challenge any grade.

When grades are released, students can engage with a Socratic tutor that understands their specific submission and the course materials. The tutor doesn't hand them answers — it asks probing questions that help them understand where their reasoning fell short.

If a student believes their grade is unfair after this process, they can submit a formal appeal directly through the same interface. The instructor receives the full conversation transcript alongside the original grading rationale, and can uphold or adjust the grade with complete context. The entire process is documented, consistent, and transparent — streamlining what has traditionally been one of the most time-consuming and opaque parts of assessment administration.

Designed for Canadian data residency.

Scaffold's infrastructure is architected to keep student data within Canadian jurisdiction. Our database, file storage, and processing pipelines are designed for deployment on Canadian-hosted infrastructure.

We build with Canada's privacy landscape in mind — PIPEDA at the federal level and Quebec's Law 25 for institutions operating in that province. For institutions subject to FIPPA, Scaffold's architecture supports the data handling requirements that provincial freedom-of-information legislation imposes on third-party service providers.

Our compliance infrastructure includes purpose-built systems for data deletion (with full audit receipts), breach record management, and granular access controls that map to institutional role hierarchies.

What we collect, and what we do with it.

Scaffold processes the minimum data necessary to deliver its service. Student submissions, rubrics, course materials, grades, and feedback are retained for the duration of the course's active use of the platform. We do not use student data for model training, advertising, or any purpose beyond delivering assessment services to the instructor and institution that authorized the processing.

Access to student data is governed by role-based permissions. Students see only their own grades and feedback. Teaching assistants see only the assessments they're assigned to. Instructors see their course. Platform administrators operate through a separate identity system that is architecturally isolated from the course-level user model.