Set up an AI tutor that answers from your course.

1. Step 1

   Open the Course Assistant for your course

   From the course page, choose Course Assistant. The first time, it opens on Settings, because nothing can be asked or added until the assistant has been created and saved.

   • The assistant belongs to one course. Each course you teach has its own settings and its own sources.
   • It starts as a draft that only you can use. Students cannot see it until you turn it on.
2. Step 2

   Name it and tell students what it is for

   Give it a name and a short description. Both are shown to students, so say plainly what it helps with and what it does not.

   • You can add up to six suggested prompts, one per line, such as "What is due this week?" or "Where is the reading for Module 3?"
   • Write prompts from questions students already email you. They set expectations for what the assistant is good at.
3. Step 3

   Choose how it teaches

   Pick Direct, Balanced or Socratic. Balanced is the default: a short explanation, then a question that checks understanding.

   • Direct explains first and then checks understanding. Socratic leads with guiding questions and hints, and explains if a student stays stuck.
   • Logistics questions such as "when is this due?" are always answered directly, whichever style you pick.
   • You can add your own guidance in plain language, for example which weekly readings to emphasize.
4. Step 4

   Choose what it can read

   On the Knowledge tab, decide which course content the assistant draws on. Its answers about your course come only from these sources.

   • From Canvas: syllabus, pages, modules and their items, assignments and rubrics, announcements, discussion prompts, quiz titles and dates, and the course calendar. Each category has its own switch.
   • Files: choose only files in published modules, all visible course files, or none. PDF, Word, PowerPoint and text files from the LMS are read up to 15 MB each.
   • Your own uploads: PDF, Word, PowerPoint, text or Markdown, up to 10 MB each. Mark an upload instructor only and students cannot retrieve it.
   • Rocketproof assignments, modules and dates are read live, so a publish change applies right away. LMS content applies at the next sync.
   • Each source shows whether it is indexing, ready or in error, so you can see what the assistant actually has before a student asks.
5. Step 5

   Set boundaries for graded work

   In "Boundaries for assessed work", write what the assistant may and may not do for your assignments, for example never suggesting a thesis statement for the Module 4 essay.

   • Platform rules always apply and your text cannot override them. The assistant is instructed never to write work a student could submit, and it cannot see grades or other students’ work.
   • Rubric explanation is off by default. If you turn it on, the assistant can explain what each Rocketproof rubric criterion values. It never shares scores.
6. Step 6

   Test it as a student, before any student does

   On the Assistant tab, choose "Test as a student". You see exactly what a student sees: only published, available content and no performance data. Test chats stay in your own history.

   • Ask a logistics question, such as a due date, and confirm the answer matches your syllabus.
   • Ask about a concept from a reading, and open the numbered source to confirm it points to the right passage.
   • Ask something your materials do not cover. The assistant is instructed to say it does not have that information and to point the student to you or the LMS.
   • Ask it to write a graded answer for you, and confirm it redirects the student to guiding questions.
   • If an answer is wrong, fix the source document or your guidance and test again.
7. Step 7

   Turn it on, and keep it current

   Choose "Turn on for students". Actively enrolled students can then use it in Rocketproof and, where your institution has set it up, from the LMS. You can turn it off at any time and your settings and sources are kept.

   • With automatic sync on, LMS content refreshes every few hours and when a launch finds the knowledge out of date. "Sync now" runs it on demand.
   • When your syllabus or a reading changes, update the source, then repeat the test questions above.

• Grades and scores, including AI-drafted scores
• Other students’ work
• Instructor private comments
• Integrity flags and the review queue
• Analytics of any kind
• Student replies in discussions
• Quiz questions and answers (only titles, descriptions and dates are read)
• Content that is unpublished, hidden or locked in the LMS

Retrieval is filtered by audience before ranking, so content a student may not see is never a candidate for an answer. For data handling, retention and subprocessors, see the security page.

• The assistant does not grade, score or judge students, and it does not detect AI use. Grading stays with you.
• It answers from the sources you give it. If a source is outdated or wrong, the answer can be too.
• Language models can still misread a passage or miss a relevant one. Citations exist so students and you can check.
• There is no dashboard of questions the assistant could not answer. Testing before launch, and fixing sources afterward, is how you find gaps.