# Canvas LMS + MentorAI Integration > Source: https://ibl.ai/resources/integrations/canvas-mentorai *Bring purpose-built AI tutoring agents into Canvas LMS — without replacing your existing workflows or surrendering your data.* Canvas LMS is the platform millions of learners and instructors rely on daily. MentorAI from ibl.ai layers intelligent, role-specific AI tutoring agents directly into that environment, meeting students where they already learn. Unlike generic chatbots bolted onto an LMS, MentorAI agents are purpose-built for education. Each agent understands course context, learner history, and institutional goals — delivering guidance that feels like a knowledgeable teaching assistant, not a search engine. Because ibl.ai runs on your infrastructure, your institution owns the agents, the data, and the outcomes. There is no vendor lock-in, and the integration is fully FERPA-compliant by design. ## Capabilities ### Contextual AI Tutoring Inside Canvas MentorAI agents launch directly within Canvas courses via LTI, giving learners instant access to a tutor that understands their current module, assignment, and progress — no context-switching required. ### Personalized Learning Pathways Agents analyze each learner's activity history, quiz performance, and engagement patterns from Canvas to adapt explanations, pacing, and practice questions to individual needs. ### AI-Generated Assignment Feedback MentorAI reviews draft submissions against instructor-defined rubrics and returns formative feedback to learners before final submission, reducing grading load and improving outcomes. ### Engagement and Mastery Analytics Session data, mastery signals, and engagement scores flow back into Canvas Analytics and the Gradebook, giving instructors a unified view of learner progress without leaving their existing workflow. ### Institution-Owned Agent Infrastructure MentorAI runs on your infrastructure. Your institution owns the agent code, conversation data, and model fine-tuning — ensuring full data sovereignty and zero vendor lock-in. ### FERPA-Compliant by Design All learner data processed by MentorAI stays within your institution's environment. No learner PII is sent to third-party AI providers without explicit institutional consent and data agreements. ## Setup ### Step 1: Provision Your MentorAI Instance (~1-2 business days) Work with ibl.ai to deploy MentorAI on your institution's cloud environment or on-premise infrastructure. Confirm domain, SSL certificates, and network access policies before proceeding. Requirements: - Signed ibl.ai service agreement - Cloud or on-premise environment meeting ibl.ai infrastructure specs - Institution IT administrator access ### Step 2: Register MentorAI as an LTI 1.3 Developer Key in Canvas (~30 minutes) In Canvas Admin, navigate to Developer Keys and create a new LTI key. Enter the MentorAI OIDC initiation URL, redirect URI, and public JWK URL provided by ibl.ai. Enable the key and note the Client ID. Requirements: - Canvas Admin account - MentorAI OIDC and JWK endpoint URLs from ibl.ai - Canvas instance on version supporting LTI 1.3 ### Step 3: Configure the Canvas REST API Connector (~45 minutes) Generate a Canvas API token with appropriate scopes (courses, enrollments, assignments, grades). Enter this token in the MentorAI admin dashboard under Integrations > Canvas to enable bidirectional data sync. Requirements: - Canvas API token with read/write scopes for courses, enrollments, and gradebook - MentorAI admin dashboard access - Allowlisted MentorAI server IP in Canvas network settings if applicable ### Step 4: Add MentorAI as an External Tool in Canvas Courses (~20 minutes per course or via global account-level deployment) In Canvas, add MentorAI as an External App using the Client ID from Step 2. Deploy it as a course navigation item, module item, or assignment submission type depending on your pedagogical model. Requirements: - Canvas course admin or instructor permissions - LTI Developer Key Client ID from Step 2 ### Step 5: Configure Agent Personas and Course Context (~1-3 hours per course) In the MentorAI dashboard, define agent personas aligned to each course or subject area. Upload syllabi, learning objectives, and rubrics so agents have accurate course context before learners interact with them. Requirements: - Course syllabi and learning objectives in PDF or text format - Instructor or instructional designer access to MentorAI dashboard ### Step 6: Test, Pilot, and Roll Out (~1-2 weeks for pilot; ongoing rollout per institution timeline) Run a pilot with a single course section. Validate that learner sessions launch correctly, data flows back to Canvas, and agent responses align with course goals. Expand to additional courses after pilot review. Requirements: - Pilot cohort of learners - Instructor and IT stakeholder review session - ibl.ai onboarding support contact ## Technical Requirements **Canvas LMS Requirements:** - Canvas LMS (Cloud or self-hosted) with LTI 1.3 support enabled - Canvas Admin account with Developer Key creation permissions - Canvas REST API access with scopes: courses:read, enrollments:read, assignments:read, submissions:read, grades:write - Canvas version released after June 2021 for full LTI 1.3 / OIDC compatibility **ibl.ai MentorAI Infrastructure:** - Cloud environment (AWS, Azure, GCP) or on-premise server meeting ibl.ai minimum specs - Minimum 8 vCPU, 32 GB RAM for standard deployment; scales with concurrent user load - PostgreSQL 14+ for relational data storage - Vector database (pgvector or Weaviate) for agent memory and semantic search - Valid SSL certificate and publicly accessible domain for LTI OIDC endpoints - Outbound network access to Canvas REST API endpoints **Security and Compliance:** - FERPA data processing agreement executed between institution and ibl.ai - All learner data must remain within institution-designated infrastructure boundaries - TLS 1.2 or higher required for all data in transit - Role-based access control configured in MentorAI admin dashboard - Audit logging enabled for all agent interactions and data access events ## FAQ **Q: Does MentorAI replace Canvas LMS or require us to migrate to a new platform?** No. MentorAI integrates directly into your existing Canvas environment via LTI 1.3. Instructors and learners continue using Canvas as normal — MentorAI appears as an embedded tool within courses. There is no migration required. **Q: Where is learner data stored when using MentorAI with Canvas?** All learner data processed by MentorAI is stored on your institution's own infrastructure. ibl.ai deploys MentorAI within your cloud or on-premise environment, so no learner PII leaves your institutional boundary. This architecture is FERPA-compliant by design. **Q: How does MentorAI know what to tutor learners on within a specific Canvas course?** When a learner launches MentorAI from within a Canvas course, the LTI integration passes course context — including the current module, assignment details, and learner activity history — to the agent. Instructors also pre-load syllabi, objectives, and rubrics into MentorAI so agents have deep course-specific knowledge. **Q: Can instructors see what learners are discussing with MentorAI agents?** Yes. Session summaries, engagement metrics, and mastery signals are written back to Canvas Analytics and the Gradebook. Instructors can also access full session logs in the MentorAI dashboard. All access is governed by role-based permissions configured by your institution. **Q: Is there a risk of academic dishonesty if learners use MentorAI for assignments?** MentorAI agents are configured to use Socratic coaching methods — guiding learners to answers through questions rather than providing direct responses. Instructors define the agent's behavior boundaries per course, and all interactions are logged for academic integrity review if needed. **Q: How long does it take to deploy the Canvas and MentorAI integration?** Infrastructure provisioning and LTI configuration typically takes 2-3 business days with ibl.ai onboarding support. A single-course pilot can be live within one week. Full institutional rollout timelines vary based on the number of courses and agent configuration complexity. **Q: Does ibl.ai support institutions that use Canvas alongside other systems like Banner or PeopleSoft?** Yes. ibl.ai is designed to integrate with the full institutional technology stack. MentorAI can consume data from SIS platforms like Banner or PeopleSoft in addition to Canvas, giving agents a richer learner profile that includes academic history and program enrollment context. **Q: What happens to our MentorAI agents if we decide to stop using ibl.ai?** Because ibl.ai deploys MentorAI on your infrastructure and you own the code and data, you are not locked into a vendor relationship. Your institution retains full access to agent configurations, interaction data, and model artifacts regardless of your commercial relationship with ibl.ai.