Authoring agent
Assignment Creator
Transforms unstructured assignment concepts into complete structured drafts with rubrics, scenarios, personas, and questioning plans.
Build with AI · Platform architecture
AI Learning Platform: Course Intelligence & Authoring
Rocketproof is an AI learning platform for higher education that unifies course intelligence, grounded student tutoring, and instructor authoring agents. Rocketproof connects to Canvas through LTI 1.3 for launch, grade passback and roster sync, syncs course content into a course-grounded assistant, and asks for confirmation before consequential agent actions. Faculty design, run and review the work.
The foundation
The course intelligence layer grounded in Canvas.
Generic foundation models hallucinate course policies and leak unreleased information. Rocketproof’s course intelligence layer anchors every response in your actual course materials.
Sync 01
Multi-source Canvas ingestion
Syncs syllabi, pages, modules, assignment rubrics, discussion prompts and course files (PDF, Word, PowerPoint, text) through a Canvas access token that your institution’s administrator provides.
Sync 02
Filter-before-rank RAG
A vital FERPA safeguard. Database queries strictly filter out unreleased assignments, draft modules, hidden faculty files, and other students' submissions before any semantic embedding search occurs.
Sync 03
Verifiable footnote citations
When an answer uses a course source, the assistant cites it with a number such as [1], and the student can open that source to check it. If the materials do not cover a question, it is instructed to say so.
Student tutoring
The Course Guide TA: Grounded and Socratic.
Provide every student with 24/7 academic support without facilitating academic dishonesty.
The Course Guide is a course-specific AI teaching assistant configured and published by the instructor of record. Unlike open-ended web chatbots, the Course Guide is instructionally bound by course ground truth and strict pedagogical guardrails.
- Answers student questions strictly from approved course readings and lectures
- Employs Socratic pedagogy on graded tasks to guide conceptual inquiry
- Is instructed not to write completed essays, solve exam problems, or supply code solutions
- Cites the course sources it uses so students can check answers for themselves
Setting one up for your course? Follow the step-by-step setup guide.
Students get immediate clarification on assignment parameters, reading arguments, and syllabus schedules without faculty email overload.
Grounded InteractionCourse Guide TA
Student: “Can you explain what Professor Davis means by ‘institutional isomorphism’ in Week 4?”
Course Guide: “According to the Week 4 reading (DiMaggio & Powell, 1983, p. 149) [1], institutional isomorphism describes the constraining process that forces one unit in a population to resemble other units facing the same environmental conditions. The author outlines three mechanisms: coercive, mimetic, and normative. Which mechanism applies to your upcoming paper topic?”
Authoring agents
Specialized authoring agents for faculty productivity.
Accelerate curriculum development with specialized AI agents designed to handle specific course authoring tasks.
Authoring agent
Course Builder
Ingests syllabus briefs and reference documents (.pdf, .docx, .pptx) to plan cohesive modules, assignments, and learning objectives as unpublished drafts.
Authoring agent
Rubric Builder
Generates detailed, multi-tier evaluation criteria, descriptors, and scoring scales aligned with departmental accreditation standards.
Authoring agent
Question Bank Builder
Synthesizes adaptive, Bloom-aligned question pools tailored to specific readings, disciplines, and difficulty bands for instructor review.
Governance architecture
The unified instructor agent Rocket with tiered governance.
Rocket is the instructor’s workspace agent. It is enabled per institution, and the ability to propose changes is a separate setting.
Tier 0 & 1
Read & Draft
Tier 0 (Read-Only): Runs read queries like overview metrics and analytics right away.
Tier 1 (Drafting): Produces a draft and applies nothing to your course.
Tier 2 & 3
Reversible & Consequential
Tier 2 (Reversible): Defined for reversible changes that need an Apply click. No current Rocket action uses it.
Tier 3 (Consequential): A confirmation card showing the impact is required to publish assignments or sync grades to the LMS.
Tier 4
Bulk-Change Guardrails
Tier 4 (Bulk): Requires typing a confirmation phrase before a bulk action such as shifting every upcoming due date in a course.
Frequently asked questions
AI learning platform questions answered.
What is an AI learning platform with course intelligence?
Course intelligence is the layer that lets an AI assistant answer from a specific course. In Rocketproof, the instructor chooses which course content the Course Assistant can read, including synced Canvas content and uploaded files. Students ask it questions, and it answers from those sources and cites the ones it uses. If the materials do not cover a question, it is instructed to say so. Language models can still make mistakes, so instructors test it before students use it.
How do AI-powered learning tools integrate with learning management systems?
Rocketproof connects to a learning management system as an LTI 1.3 tool, with deep linking, grade passback and roster sync. Canvas is the platform verified end to end, from registration through a graded launch. Grades pass back to the Canvas gradebook after the instructor reviews and releases them, where passback is configured. Instructor comments do not pass back. Course content for the Course Assistant syncs separately, using a Canvas access token that your institution’s administrator provides.
How does the Course Guide handle graded work?
On graded assignments, quizzes, exams and defenses, the Course Guide is instructed to guide students with questions, an outline and pointers to course material, and to decline to write an essay, code or answer a student could submit. Instructors can add boundaries for specific assignments. These are instructions to a language model and are not a guarantee, so test the assignments you care about before students use it. Separately, content a student may not see, such as unpublished assignments and other students’ work, is filtered out before retrieval, so it is never a candidate for an answer.
What is tiered governance for AI agents?
Tiered governance sorts the actions an AI agent can take by risk. Rocket, the instructor agent, uses five tiers. Tier 0 actions read data and run immediately. Tier 1 actions draft something and apply nothing to the course. Tier 2 is defined for reversible changes that need an Apply click, and no current Rocket action uses it. Tier 3 actions with educational consequences, such as publishing an assignment or syncing a grade to the LMS, always show a confirmation card with their impact. Tier 4 actions change many records at once, such as shifting every upcoming due date, and require typing a confirmation phrase. Rocket is enabled per institution, and its actions are a separate setting.
Build with AI safely in your courses.
Request a pilot, or try our interactive demo to experience course intelligence in action.