Illustration of a student asking a question to iBot, which is reading from Moodle content and providing an answer with citations.
AI teaching assistant for Moodle

Ask your course anything.
Get answers it can prove.

iBot reads your institution's own Moodle content and answers, tutors, quizzes, and tracks every student against it — with citations back to the source, hosted and managed end-to-end by iLite.

See it on your campus How grounding works
iBot · Introduction to Statistics
How is the final exam weighted?

Your final exam counts for 40% of your overall grade. The remaining 60% is split across weekly quizzes (20%) and the group project (40%), as set out in the assessment breakdown released in Week 2.

Assessment Guide Syllabus — Week 2
Two course pages disagree on this figure — iBot flagged the conflict and answered from the more recently updated one.
Icon for 2 sources2 sourcesMoodle content + uploaded docs, merged into one search
Icon for 3 teaching styles3 teaching stylesDirect, Socratic, and Scaffolded — selectable per course
Icon for 14 skills14 skillsTracked per learner against a configurable competency framework
Icon for 100% managed100% managedHosted and run end-to-end by iLite — nothing for your institution to host
Illustration of iBot's neurosymbolic router deciding how to answer a question.
The architecture

A neurosymbolic router decides how to answer — before it answers

Every question passes through a layer that combines a trained neural classifier with deterministic symbolic rules. The model proposes an intent and a confidence score; the rules decide whether that read is trustworthy enough to act on. When it isn't, iBot asks a clarifying question instead of guessing.

Step 1

Symbolic pre-filter

Deterministic rules catch greetings and obviously out-of-scope chatter before any model runs.

Step 2

Neural classification

A trained router scores the question against every possible route and returns a confidence value.

Step 3

Symbolic decision gate

Formal rules check a confidence floor and an ambiguity margin between the top two candidate routes.

Step 4

Dispatch

A confident, unambiguous read is sent to one of three destinations, below.

Platform questions

Moodle-style FAQ

“Why won't my quiz submit?”

Matched against a Moodle platform knowledge base — answered instantly, no retrieval or generation needed.

Personal questions

Learning path & records

“What's my grade in Week 4?”

Routed to the student's own live Moodle data — grades, deadlines, enrolments, competency profile — never to course content search.

Content questions

Direct course-content answers

“Explain the central limit theorem.”

Sent through full retrieval, reranking, and cited generation against the course's own material.

Illustration of a student interacting with iBot in a Moodle course.
What's live today

Talk to your course

Everything below is shipped and running on real Moodle courses.

Icon for Grounded, cited answers

Grounded, cited answers

Retrieves across Moodle pages, books, and uploaded documents in one pass. Every claim cites its source; contradicting content is flagged, not guessed at.

Icon for Instant practice quizzes

Instant practice quizzes

Ask “quiz me on this” on any course page and iBot builds a multiple-choice quiz on the spot from that page's content, with instant right/wrong feedback.

Icon for “Build me a course on…”

“Build me a course on…”

A short interview — prior knowledge, depth, objectives — then a full outline generated from real course material and saved straight to Moodle.

Icon for Multilingual by default

Multilingual by default

Chat responses and full page content translate on demand, so a single course serves cohorts studying in more than one language.

Response modes

One assistant, three ways to teach

The same question can be answered three different ways — configurable per course, so the assistant matches how an instructor actually wants students to learn.

Direct

Answers straight, with citations

The default mode. Best for fast, factual lookups where the student just needs the answer.

“The final exam is 40% of your grade, per the Week 2 assessment guide.”

Socratic

Answers with a guiding question

Prompts the student to reason toward the answer instead of handing it over.

“What happens to current if resistance increases but voltage stays fixed?”

Scaffolding

Breaks it into steps, with hints

Splits a hard question into smaller pieces, revealing more only as the student engages.

“Let's start with what Ohm's Law tells us about voltage and current…”

What's live today

See every learner clearly

One dashboard shell, five views, all reachable without leaving the page a student is already on.

Icon for Home

Home

Deadlines, competency highlights, and recommendations at a glance

Icon for Chat

Chat

Grounded, cited conversation with the course, in real time

Icon for Predictive

Predictive

Full competency profile, skill trends, and recommendations

Icon for Notes

Notes

Pin-style annotations captured while reading, searchable later

Icon for History

History

Every past conversation, saved and revisitable

The Predictive tab

From one question to a full learning picture

Built on nightly-refreshed competency data and real quiz activity — not a one-time survey.

Competency profile

Ranked strengths and gaps across a configurable skills framework, rebuilt nightly from real activity.

Smart recommendations

Course suggestions that blend stated interest with measured skill gaps — each one explained, not just listed.

Performance trends

Per-skill accuracy from graded quiz activity, with confidence flags and improving / declining / stable trend badges.

AI-generated course reviews

A short, pre-generated advisory review for each course, written for what a struggling or advanced student needs to know.

Upcoming deadlines

Pulled live from the student's own Moodle calendar — no separate login, no manual entry.

Page-level analytics

Every course page gets a readability score, and iBot quietly tracks real engagement — scroll depth, pauses, re-reading — to sense when a student is stuck before they ask.

Illustration of iBot's hosted, multi-tenant infrastructure.
What's live today

Built to scale

The infrastructure choices that matter to an IT director, not just to a student.

Hosted & managed by iLite

The language model runs on infrastructure iLite hosts and manages for you — nothing for your institution to run, and no student conversation is sent to a public LLM API.

Multi-tenant by design

One stack can serve several institutions in isolation — each with its own data, license, and content.

JWT-secured sessions

Audience-validated tokens on every real-time chat connection — not just on the initial page load.

Nightly data sync

Grades, deadlines, and skill profiles refresh automatically every night — no manual exports, ever.

Illustration of the iBot roadmap.
In development

Where iBot is headed next

Active investments, not commitments on a date. Each builds directly on data iBot is already collecting today.

Adaptive tutoring, everywhere

Beyond Q&A

Extending Socratic and Scaffolding modes from course-content answers into quizzes and course building, with an instructor-facing control to choose the style per class.

Mastery modeling

From snapshots to a live picture

Moving beyond nightly competency profiles toward a continuously updated estimate of what each student has and hasn't mastered, once per-question tagging or a larger skills framework unlocks it.

Trend history

Momentum, not just position

Weekly skill-accuracy snapshots alongside current standing, so advisors can see who's improving, plateauing, or falling behind — early.

Illustration representing LTI 1.3 integration.
LTI 1.3

LTI 1.3 integration

Built on the LTI 1.3 standard, with role and course context read directly from Moodle's signed launch — so instructors and students land in different views of the exact same activity, with no extra logins, config, or spoofable parameters.

Scenarios

Scenario modes

Use scenario modes to create interactive learning experiences that adapt to student choices and actions within the course.

Grades

Gradebook integration

Every scenario attempt writes a score straight to your Moodle gradebook through the LTI Assignment & Grades Service — turn-count and participation based, so the number in your gradebook stays predictable and easy to defend.

Teacher review

Teacher review

Launching the same scenario activity as an instructor swaps in a review view instead of the chat: every attempt at that specific activity, with mode, turn count, status, score, and when it happened — plus a secure drill-down into the full transcript, automatically scoped to the instructor's own course.

Under the hood Vector search· Managed LLM inference· Neurosymbolic routing· Moodle Web Services API· Real-time secure chat

Bring iBot to your Moodle instance

Talk to us about a pilot on one course, one department, or your whole institution.

Get in touch →