Feedback is most valuable when students can act on it. Student Tutor gives them a place to ask what a comment means, retry a missed step, or practice the same skill while the assignment is still fresh.
You should see
Students leave the feedback page with a clearer understanding of what went wrong and one concrete way to improve.
Key points
- It helps students understand the why behind lost points instead of only seeing a score.
- It turns feedback into a next action: explain, fix, or practice.
- It keeps help tied to the student's own work, assignment, grade, and feedback.
- It reduces the gap between receiving feedback and knowing how to improve.
Students start from the same feedback link they already use. The tutor appears as a second tab beside the feedback view.
Step by step
- The teacher shares feedback through LMS publish, email feedback, or a copied secure feedback link.
- The student opens that secure feedback link from the LMS, email, or copied URL.
- The page shows the returned grade, download action, and the Feedback / Tutor pivot.
- The student reviews the Feedback tab to see question-level comments and scores.
- The student switches to Tutor when they want help turning that feedback into next steps.
- The student can use suggested prompt chips or type a custom question in their own words.
You should see
The student can move from reading feedback to asking a targeted follow-up question in the same feedback page.
Keep in mind
- Students do not need to enter the teacher dashboard.
- The tutor stays scoped to one assignment and one secure feedback link.
- The tutor is meant to support learning from returned work, not replace teacher review.
Student Tutor works best for targeted follow-up after feedback has been returned. It is not a general homework bot.
Key points
- Ask why points were lost on a specific question or feedback comment.
- Ask to walk through a missed step without just being handed the answer.
- Ask for a similar practice problem that tests the same concept.
- Ask for a simpler explanation when the teacher feedback feels unclear.
- Ask what to check next before revising or studying.
Keep in mind
Students should not use the tutor to challenge a grade, request the answer key, or ask unrelated homework questions. Those should go to the teacher or the appropriate class support channel.
The tutor is personalized to the student response, but protected teacher materials stay internal.
Key points
- Tutor context includes the student name, assignment name, course context, score, published feedback, and submitted work when available.
- The answer key and rubric can guide the tutor internally, but the tutor should not quote or expose them to students.
- Teacher AI instructions are included when available, as long as they do not override answer-key protection or safety rules.
- The tutor should guide reasoning, ask helpful follow-up questions, and generate original practice problems when useful.
Keep in mind
The tutor does not regrade work, change scores, or confirm rubric point values. Students should still contact the teacher if they believe feedback or scoring is wrong.
The chat is designed to feel responsive while still protecting the student from stuck or overly long model responses.
Step by step
- When a tutor session starts, Classwise creates or reloads the active session for that feedback link.
- The opening message and each tutor reply can return three suggested follow-up prompts.
- While a response is processing, students can keep typing and can stop the in-progress response.
- If the model times out or returns an invalid response, the student sees a short fallback message and can try again.
You should see
The chat remains usable even when a model response is slow, stopped, or replaced with a fallback message.
Keep in mind
- Tutor sessions and messages are stored so the conversation can reload later.
- PostHog events track tutor load, send, stop, success, model-error, and failure states.
- The current timeout is designed to prevent a stuck chat experience.
Common blockers
- The feedback link must point to published feedback and a matching submission.
- Model timeouts can still happen, but they should show a recoverable fallback instead of blanking the tutor UI.