# Tutoring Agent

> Source: https://ibl.ai/resources/agents/tutoring-agent


*Adaptive, curriculum-grounded tutoring an institution owns — available to every student, with student data that stays on the institution's infrastructure.*

The Tutoring Agent is an autonomous AI agent that gives every student adaptive, one-on-one tutoring grounded in the institution's own curriculum and course materials.

It meets students where they are, works through problems step by step, tracks mastery, and surfaces struggling students to staff, instead of help being limited to office hours and a waitlist.

This is an institutional tutoring system the school or university owns, not a consumer app. It deploys on the institution's infrastructure, with FERPA and COPPA control and student data that never leaves the institution.

## Agent vs. Chatbot

A homework chatbot gives an answer. The Tutoring Agent teaches — grounded in your curriculum, adapting to the student, tracking mastery, and flagging who needs human help — autonomously, for every learner.

| Dimension | Chatbot | Agent |
|-----------|---------|-------|
| Execution | Hands back an answer to a question | Guides the student through the concept and checks for mastery before moving on |
| Grounding | Answers from the open web | Teaches from the institution's own curriculum, course materials, and standards |
| Memory | Stateless — forgets the student | Tracks each student's progress, gaps, and mastery across sessions |
| Tools & APIs | Cannot reach the LMS or SIS | Connects to Canvas, Moodle, and PowerSchool to align with courses and surface at-risk students |
| Data Control | Student data leaves to a vendor cloud | Runs on the institution's infrastructure; student data never leaves it |
| Model Flexibility | Locked to one vendor's model | Model-agnostic — Claude, GPT, Gemini, Llama, Mistral, or a fine-tuned model |
| Safety & Compliance | Generic, age-blind responses | Grade-band-appropriate guardrails, FERPA/COPPA control, and a complete audit trail |
| Autonomy | Reacts to each prompt | Runs a continuous teach-assess-adapt-escalate cycle for every student |

## Core Capabilities

### Curriculum-Grounded Tutoring

Tutors from the institution's own course materials and standards, so guidance matches what is actually taught.

*Autonomous action:* Pulls the relevant unit and objectives for a student's course and tutors against them, not a generic syllabus.

### Adaptive, Step-by-Step Instruction

Adjusts to each student's level, working through problems with hints and checks rather than handing over answers.

*Autonomous action:* Detects where a student is stuck, adapts the explanation, and confirms understanding before advancing.

### Mastery & Progress Tracking

Tracks each student's mastery of concepts and learning gaps across sessions.

*Autonomous action:* Maintains a per-student mastery model and recommends the next best activity automatically.

### At-Risk Student Surfacing

Identifies students struggling or disengaging and flags them for human intervention.

*Autonomous action:* Alerts an advisor or instructor when a student's signals indicate they need a human, with the context attached.

### Grade-Band-Appropriate Guardrails

Applies age-appropriate moderation and academic-integrity guidance by grade band.

*Autonomous action:* Tunes responses and guardrails to the student's grade band and the institution's academic-integrity policy.

## How It Works

1. **Receive — Understand the Student:** The agent identifies the student's course, level, and current objective from the LMS and prior session history.
2. **Reason — Diagnose the Gap:** It diagnoses where the student is struggling and selects an approach grounded in the institution's curriculum.
3. **Act — Teach and Check:** The agent tutors step by step, gives hints rather than answers, and checks for understanding before moving on.
4. **Evaluate — Track Mastery:** It updates the student's mastery model, recommends next steps, and flags at-risk students to staff.
5. **Report — Inform Instructors:** The agent reports progress and common misconceptions to instructors and logs activity to an audit trail.

## ROI & Impact

| Metric | Value | Description |
|--------|-------|-------------|
| Tutoring Access | every student | Extends one-on-one tutoring to every student, not just those who reach office hours. |
| Retention & Completion Lift | measurable | Always-available, curriculum-grounded support improves retention and course completion signals. |
| Cost vs. Per-Seat Tutoring Apps | ~10x cheaper | Enterprise-wide flat-fee licensing eliminates per-student tutoring SaaS pricing. |
| Staff Time Refocused | high | Routine tutoring is automated so staff focus on students who need human intervention. |
| Data Control | complete | Student data stays on the institution's infrastructure, simplifying FERPA and COPPA compliance. |

## FAQ

**Q: How is this different from a consumer AI tutor app?**

Consumer tutor apps run in a vendor cloud and answer from the open web. The Tutoring Agent is owned by the institution, tutors from your own curriculum, runs on your infrastructure, and keeps student data under your control with grade-band guardrails.

**Q: Is it FERPA and COPPA compliant?**

Compliance is a property of deployment. The agent runs on the institution's infrastructure, so student data does not leave the institution, with access controls, grade-band moderation, and a complete audit trail — the cleanest basis for FERPA and COPPA.

**Q: Does it just give students answers?**

No. It tutors step by step with hints and checks for understanding, applying the institution's academic-integrity policy, rather than handing over answers. The goal is mastery, not completion.

**Q: Which LMS and SIS does it integrate with?**

It integrates with Canvas, Blackboard, Moodle, D2L Brightspace, PowerSchool, Clever, and ClassLink, aligning tutoring with courses and surfacing progress and at-risk signals to staff.

**Q: Can instructors see how students are doing?**

Yes. It tracks per-student mastery and common misconceptions, surfaces at-risk students for intervention, and reports progress to instructors and advisors.

**Q: Do we own the source code?**

Yes. ibl.ai delivers the complete source code, so the institution can audit, modify, and operate the tutoring system permanently, independent of ibl.ai's pricing or roadmap.