Roxy Tutor: Personal AI Tutoring Built on Mastery

Most AI tutoring starts with a message box. The learner explains the subject, repeats the goal, pastes the same notes and hopes the next answer fits.

Roxy Tutor starts with a relationship instead. The learner sets the subject, current level, curriculum, goals, learning needs and weekly schedule. Those details remain attached to that tutor, together with the trusted sources and coaching history that make later sessions useful.

The other important difference sits underneath the conversation: native Mastery, not tutor chat, decides learning progress.

One product, two clear responsibilities

Roxy Tutor and Mastery work together, but they do different jobs.

Roxy Tutor owns	Native Mastery owns
The private tutor relationship	Diagnostic evidence
Subject, goals and learner context	Modules and learning objectives
Curriculum, level and schedule	The next objective
Bound notes and knowledge sources	Checks, grading and retry
Coaching history and progress context	Spaced review and completion

That boundary matters. A friendly conversation is not proof that a learner understands a topic. Mastery needs evidence from an attempt before it advances an objective. Roxy Tutor can explain, encourage and keep context, but it does not silently turn chat activity into a passing grade.

What a normal Roxy Tutor session looks like

A learner should not need to understand the architecture to use the product. The everyday flow is straightforward:

1. Create the tutor. Choose a subject and add the learner's level, curriculum, goals, needs and schedule.
2. Add trusted material. Bind class notes, worksheets, readings, slides or other knowledge sources to the tutor.
3. Start the diagnostic. Roxy Tutor opens the canonical Mastery session and gathers enough evidence before activating a path.
4. Work on one next objective. The learner gets an explanation, a guided attempt or a focused check tied to the active objective.
5. Recover after a miss. An incorrect or unsure answer leads to explanation and retry. It does not create false progress.
6. Return for review. The learner can reload the tutor, continue the same relationship and complete reviews when Mastery schedules them.

The result feels personal without becoming vague. The tutor remembers why the learner is here. Mastery keeps the learning state honest.

Personal context that survives the first session

Personalisation should be more than adding a name to a prompt. Roxy Tutor keeps structured details that can change how teaching support is framed:

• country and education system;
• state or province;
• curriculum or exam framework;
• current year or level;
• the learner's goals and current difficulties;
• learning needs the tutor should understand;
• the days and amount of time available for study.

These details can be reviewed and edited. They are coaching context, not a permanent label on the learner.

Source-grounded tutoring

A personal AI tutor is more useful when it can work from the material the learner is actually using. Roxy Tutor can be bound to knowledge sources so an explanation stays close to class notes, readings, worksheets, slides and drafts.

The binding is controlled by the tutor relationship. A learner should not be able to accidentally swap in an unrelated source during a protected Mastery session and change the basis of the lesson.

This also makes follow-up work easier. The learner can ask for another explanation, open a related study tool or revisit the original source without rebuilding the context from scratch.

What parents and learners should expect

Roxy Tutor is study support. It does not replace a teacher, school policy, marking criteria or professional advice.

It should make several things clearer:

• what the learner is working toward;
• what evidence Mastery has collected;
• why a particular objective is next;
• whether today's activity is required study, optional catch-up or review;
• what remains before the path can be called complete.

That clarity is more valuable than a stream of confident answers. A useful tutor should know when to explain, when to ask, when to retry and when the evidence is not strong enough yet.

Start with one subject

The best first test is a real subject with real material. Create a Roxy Tutor, add the learner's current goal and bind the notes or worksheet they are already using.

For the broader design argument, read Why Context-Aware AI Tutoring Beats Generic Chatbots. For families working from assessment evidence, AI NAPLAN Insights offers a separate public report-analysis flow.