Questions worth asking.

Because that is where an AI tutor can be genuinely good, and honest. Maths and the sciences have checkable answers, so Saga can withhold the solution, coach a student to it, and verify objectively that the skill has stuck. The whole mastery loop depends on being able to tell "right" from "not yet".

Open-ended writing is different. Automated essay scoring is not reliable or fair enough to stake a child's learning on, and it is known to disadvantage some students, including those writing in an additional language. We would rather do a few subjects properly than pretend to grade something we can't. In open subjects the right role for AI is a coach, not a judge, and that is a different product.

Engagement is necessary but it is not the goal; learning is. A tool can be delightful and teach nothing. So every Saga lesson has a check before and a check after, and a step only counts as mastered when the student can redo it unaided. That turns "they seemed to enjoy it" into "here is what they can now do that they couldn't before".

Measuring gain also keeps us honest. It is how a teacher can see whether Saga is working for a particular class, and it is how we decide what to improve.

It is the part that makes one-to-one tutoring so powerful: adapting to the student in front of you. When a student is stuck, Saga works out why, tracing the struggle back through a prerequisite map to the foundation that is actually missing, then shores that up with a simpler example before returning to the problem. Students who race ahead get optional challenge problems instead.

It adapts the route, never the destination. The student still states every answer themselves. See adaptive tutoring in detail.

No, and this is deliberate. On a problem that is meant to be worked through, Saga withholds the answer and coaches the next step with graded hints. A deterministic guard sits behind the model: if a draft reply would state the answer before the student has, it is caught and rewritten. The student is always the one who says the answer first.

Saga covers every KS3 and KS4 maths and science objective of the English National Curriculum, and it teaches the IB Middle Years Programme for international schools. Lessons are sequenced to a prerequisite map reviewed by subject experts, and a teacher can see exactly which objectives are covered and how they build. See the curriculum.

Safeguarding is built in, not bolted on. There is a clear, tested path from a worried child to a trusted adult at the school, and the tutor holds firm conduct boundaries. To a teacher, a flag appears as a marker only, that something was raised and the right adult notified, never the content. How we keep children safe.

Yes, by design. Students are pseudonymous, we minimise the data we hold, and the student-facing view carries no grades or attainment scores, only formative guidance. By default, text-to-speech runs on the device itself, so nothing is sent away to be spoken; a school can opt into a more natural cloud voice, which sends only the tutor's message text to the speech service, never a student's voice or data. We work to the data rules of your region.

No. Saga is a one-to-one supplement to good teaching, not a substitute for it. The teacher decides what is assigned and stays fully in control; Saga gives them time and insight back, coverage, progress, and a coaching view of each student. See the insight teachers get.

Every lesson is grounded in authored content with a written solution, so the tutor coaches from a known-correct source rather than improvising. The answer-leak guard, the separate marking check, and continuous monitoring for leaks are all there to keep the tutoring accurate and honest, and we are upfront about the limits.

Broadly, students aged 11 to 16: KS3 and KS4, and MYP for international schools. Saga pitches its language to the year group, so the concept is the challenge and the reading level isn't, and it can read every message aloud for students who learn best by listening.