The AI teaching assistant your faculty controls.
EdPilot grounds every answer in your actual course materials, inside guardrails professors set. Students get a tutor that knows the class. Universities get AI adoption on their terms.
University-led evaluation · Set up in minutes · No IT project required
Student Workspace
The answer stays inside the course.
- The AI only knows what the professor uploadsSyllabus, slides, and rubrics become the assistant's entire world, not the open internet.
- Every answer cites the courseStudents can click through to the exact slide or page, so trust never depends on vibes.
- Professors set rules the AI can't breakIntegrity mode, citation policy, and assessment boundaries are enforced on every reply.
Students are already using AI on your courses.
Right now that happens through generic chatbots: outside your policies, outside your visibility, and often wrong about your class. The real question is whether your institution has any say in how.
Ungoverned
Generic chatbots follow their own rules, not your syllabus, your honor code, or your assessment policy.
Invisible
Faculty get no signal about what students ask, where they struggle, or how AI is shaping their learning.
Unaccountable
Answers come from the open internet with no citations, confidently wrong about your course and impossible to check.
- of undergraduates already use AI — source: Higher Education Policy Institute, Student Generative AI Survey 2026
- 95%
- of US students used generative AI for coursework last year — source: Inside Higher Ed, Student Voice survey, 2025
- 85%
- of teens say AI cheating is a regular occurrence at their school — source: Pew Research Center, 2026
- 59%
- of faculty can reliably identify AI-generated work — source: Frontiers in Education, 2025
- 29%
of undergraduates already use AI
SourceHigher Education Policy Institute, Student Generative AI Survey 2026
of US students used generative AI for coursework last year
SourceInside Higher Ed, Student Voice survey, 2025
of teens say AI cheating is a regular occurrence at their school
SourcePew Research Center, 2026
of faculty can reliably identify AI-generated work
SourceFrontiers in Education, 2025
The question is no longer whether students use AI on your courses. It's whether the institution can see it, shape it, and answer for it.
Answers from the open internet. No citation, no syllabus, no record the institution can review.
Answers from the professor's own materials, cited to the page, inside the guardrails the course set.
From syllabus to student support in an afternoon.
No IT project, no migration. A professor uploads materials, sets the rules, and shares a link. The course model does the rest.
Upload course materials
Syllabi, lectures, readings, rubrics, policies, and assignments become the source of truth.
Configure guardrails
Faculty choose citation rules, assessment boundaries, tone, and what the assistant can answer.
Students ask safely
The assistant gives grounded explanations, practice prompts, and hints without completing work.
Faculty see patterns
Confusion, misuse attempts, and concept gaps become visible before the next assessment.
- To register your university
- ~2 min
- From sign-up to teaching faculty
- Same week
- Student records used to train public models
- 0
- Answers grounded in your course materials
- 100%
To register your university
From sign-up to teaching faculty
Student records used to train public models
Answers grounded in your course materials
One campus decision. Three clear experiences.
The university owns the evaluation. Each role gets a focused experience inside the same governed academic boundary.
Administrators
A governed path for campus AI adoption.
Pilot course-grounded AI with privacy posture, rollout controls, and faculty ownership built into the experience.
Professors
Fewer repetitive questions, more useful signals.
Set the knowledge boundary, review source-backed answers, and see where students are struggling before office hours fill up.
Why divide by n−1 and not n?
The sample mean is estimated from the same data, so one degree of freedom is already spent.
Lecture 4, p. 12Students
24/7 help that speaks the language of the class.
Get explanations, practice prompts, and citations from the actual syllabus, slides, readings, and rubrics.
Built for higher education, not generic chat.
Other AI tools are built for everyone. EdPilot is scoped to a course, governed by faculty, and designed for institutional review.
Course-specific by default
Every answer is grounded in uploaded course materials, not a generic web-scale guess.
Faculty-controlled
Instructors define what the AI knows, how it responds, and where it stops.
Integrity-first
Designed to guide students toward understanding instead of completing assessed work.
Insight-rich
Show where students are confused before confusion becomes an exam result.
AI Teaching Assistant
A teaching assistant that answers student questions at 2am using your materials, your terminology, and your standards.
- Cites your readings instead of random internet sources.
- Flags misconceptions before they reach the exam.
- Follows the guardrails and knowledge boundary you set.
- Separates fast pilots from responsible institutional rollout.
Meets your campus where it already works.
EdPilot runs in the browser and connects to the tools your courses already live in, starting with Canvas.
Canvas LMS
Sync courses, rosters, assignments, and due dates directly from Canvas, so the assistant knows what's due and when.
Works with or without an LMS
Fully browser-based. No installation, no plugin approval, no migration. Professors share a link and students are in.
LTI 1.3 embedding
Deeper embedding inside Canvas, Moodle, and Blackboard through the LTI standard is on the roadmap.
Prepared for the questions buyers actually ask.
EdPilot makes the academic, privacy, and implementation posture visible before a pilot turns into a procurement surprise.
FERPA & student data
Handling built around institution-bound course and student data. Student records never train public models.
Data boundaries
Course materials, interactions, and deployments are scoped by institution and course, encrypted in transit and at rest.
Compliance status
SOC 2 Type II audit in progress. Procurement-ready notes cover data handling, LMS status, retention, and rollout.
WCAG 2.2 AA accessibility
Interaction patterns and content flows are designed and reviewed against WCAG 2.2 AA.
Faculty ownership
Course control stays with the instructor instead of moving into a generic AI layer.
Integrity controls
Assessment and homework requests can be routed toward hints, practice, or refusal states.
What people ask before they pilot.
How is EdPilot different from ChatGPT?
Do professors control what the AI can and can't do?
How long does setup take?
What does it cost to try?
How is student data handled?
See EdPilot on your course materials.
Plan a university-led pilot with real syllabus content. Set up in minutes, evaluated on your terms.