Grade open-ended work against teacher-defined criteria
Assess essays, spoken interviews, photos, and presentations with a visible criterion-level first pass and a human final decision.
Student response
Each step stays editable, visible, and connected
These product views use example data to show what the teacher actually sees—not an abstract feature diagram.
- Step 01
Task + rubric
Build an editable exercise pool from the actual lesson context, then select the tasks, response formats, difficulty, and points that fit this learner.
practiceTask + rubric
Lesson contextInterview evidence and decision risks01 Explain your recommendationExtended answer · 10 pointsIncluded02 Prioritise the evidence gapsScenario · 8 pointsIncluded03 Record a two-minute defenceSpoken response · 12 pointsReview - Step 02
Student response
The original written, spoken, visual, or presentation response stays beside its transcript and assignment context, so evidence is never reduced to a score alone.
responseStudent response
AssignmentDefend the decision and name the largest uncertainty.▶02:14Student response“The strongest evidence supports option B, but the sample is still too narrow…”
Transcript ready - Step 03
AI first pass
AI creates a criterion-level first pass with a score, evidence, and uncertainty for the teacher to inspect—not an invisible final decision.
practiceAI first pass
Lesson contextInterview evidence and decision risks01 Explain your recommendationExtended answer · 10 pointsIncluded02 Prioritise the evidence gapsScenario · 8 pointsIncluded03 Record a two-minute defenceSpoken response · 12 pointsReview - Step 04
Evidence by criterion
Review the student’s essay, audio, photo, or presentation—not only an automated summary.
responseEvidence by criterion
AssignmentDefend the decision and name the largest uncertainty.▶02:14Student response“The strongest evidence supports option B, but the sample is still too narrow…”
Transcript ready - Step 05
Teacher review
The teacher compares the first pass with the original work, changes the score where needed, writes final feedback, and decides what the learner receives.
reviewTeacher review
AI first pass84%Ready for teacher review- Uses evidence4 / 5
- Explains limitations5 / 5
- Recommendation clarity3 / 5
Teacher feedbackGood reasoning. Add one piece of evidence that could disprove your recommendation. - Step 06
Final feedback
The teacher compares the first pass with the original work, changes the score where needed, writes final feedback, and decides what the learner receives.
reviewFinal feedback
AI first pass84%Ready for teacher review- Uses evidence4 / 5
- Explains limitations5 / 5
- Recommendation clarity3 / 5
Teacher feedbackGood reasoning. Add one piece of evidence that could disprove your recommendation.
The richest answers are the slowest to review
Meaningful work rarely fits an exact answer key, and two partially correct learners may need different feedback.
Define evidence before scoring
Provide the task, expected answer, assessment criteria, weights, and what observable evidence supports each level.
Receive criterion-level feedback
See what evidence was found, what is missing, and how the suggested score relates to each criterion.
Inspect the original submission
Review the student’s essay, audio, photo, or presentation—not only an automated summary.
Change the result and escalate uncertainty
Teachers can override the score and should manually handle uncertain, high-stakes, sensitive, or unsupported cases.
Common questions
What source formats can I use with AI grading?
Meaningful work rarely fits an exact answer key, and two partially correct learners may need different feedback. Only use source material you are authorised to process, and remove unnecessary personal or sensitive information.
Does AI grading use my own material as context?
Meaningful work rarely fits an exact answer key, and two partially correct learners may need different feedback. The selected description, document, recording, lesson, or response remains the grounding context for the requested workflow.
Can I edit the output from AI grading?
Provide the task, expected answer, assessment criteria, weights, and what observable evidence supports each level. Generated modules, lessons, exercises, criteria, scores, and feedback are saved as editable product objects rather than a block of text to copy elsewhere.
What happens to files and recordings uploaded for AI grading?
Only upload material you are authorised to process. Keep personal data to the minimum needed for the teaching task, review the result before sharing it, and follow your organisation’s retention and consent policy.
Can a teacher override the result from AI grading?
See what evidence was found, what is missing, and how the suggested score relates to each criterion. AI can prepare or assess a first pass, but the educator controls source material, assignment, publication, the final score, and student-facing feedback.
Which subjects and languages work with AI grading?
See what evidence was found, what is missing, and how the suggested score relates to each criterion. Support depends on the source quality and requested subject, so educators should review terminology, notation, cultural context, and assessment expectations before use.
How is AI grading priced?
AI usage is metered with the same SubSchool AI balance used by course, lesson, homework, and assessment workflows. You can keep drafts private and review the expected operation before publishing or assigning the result.
Can I use AI grading without publishing a public course?
Review the student’s essay, audio, photo, or presentation—not only an automated summary. You can keep work private as a draft, test the complete flow with example data, and publish or assign it only after review.
Turn the page into a real teaching workflow
Keep the result editable, connect it to learners, and preserve teacher review.