Most books about AI and education are written for policymakers, researchers, or enthusiasts. This one is written for the parent sitting at the kitchen table, watching their Year 8 child open a chat window instead of a textbook, and quietly wondering: is this actually any good?
AI Tutors for Key Stage 3 — published in 2026 by Innovatorly Ltd, with a Simplified Chinese edition (《英国中学的AI私教》) released simultaneously — is a full-length illustrated guide for exactly that parent. Its author, Duke Harewood, built the tutoring system behind aitutors.me before writing a single word of the book. The result is something unusual: a guide that gives the entire method away.
What the book actually is
At roughly 40 illustrated chapters — each diagram a clear, simple line drawing rather than a decorative flourish — the book covers six structured parts. Part I sets the scene: why AI tutoring has arrived at the secondary level and what it can realistically do. Part II, the longest and arguably the most distinctive, is a plain-English AI-literacy primer aimed squarely at non-technical parents. It explains what a large language model is, how a harness and an agent relate, what MCP means, and why skills, slash commands, workflows, hooks, plugins, and memories matter. It also explains why each tutor has its own personality rather than being a single generic chatbot.
Parts III through VI move from the technology into the teaching. Part III introduces the faculty. Part IV documents the methods in detail — the four-level hint ladder, the weekly energy system, the mastery loop, and the safeguarding protocols. Part V is a practical playbook for parents: how to use the tutors alongside school, how to talk to your child about AI, and what to watch for. Part VI covers the evidence base and curriculum alignment, and closes with an appendix addressed to educators who want to understand what their students may already be using at home.
Meet the eight tutors
The book's faculty is specific and deliberate. There is a Mentor — a wellbeing-first guide whose job is to check in on how a student is feeling before routing them anywhere near a subject. Then there are seven subject specialists: Professor Pi (mathematics), Professor Quill (English), Professor Darwin (biology), Professor Curie (chemistry), Professor Newton (physics), Professor Harari (history), and Professor Mercator (geography).
Each tutor has a distinct voice and set of tools. Professor Pi, for example, uses a four-level hint ladder that starts with a nudge and works toward a full worked example — but only if the student still cannot proceed after three guided attempts. Professor Harari approaches history through source analysis rather than rote recall. Mentor, meanwhile, never delivers subject content at all; its role is to hold the relationship and notice when something is wrong.
If you want to understand why each tutor is built the way it is — including the pedagogical reasoning behind personality and tone — Part III is the section to read first.
The one rule
The book could have been subtitled The One Rule, because everything in it descends from a single principle: a good tutor never gives the final answer. It hands the last step back to the student.
This is the Socratic method applied with consistency. Not because it sounds virtuous, but because it is the only approach that builds the kind of knowledge that survives a closed-book exam. A student who has been walked through forty solved problems has not learned to solve problems; they have learned to watch. A student who has been nudged up a four-level hint ladder forty times has built the neural pathways that belong to them.
The difficulty is that consistency is hard for humans and easy for machines. A tired parent at 9pm will give the answer. A well-designed AI tutor will not. The book explores the Socratic method in AI tutoring in depth, including why it occasionally frustrates students and why that frustration is a sign the method is working.
Why a book, and why bilingual
The honest answer to the first question is that the aitutors.me service does not explain itself. Parents see their child talking to a tutor, they see homework getting done, they occasionally see a strange message like "I'm not going to give you the answer to part (c) — try writing the equation from scratch and I'll check your reasoning." They do not see the method behind it.
The book is the explanation. It is an attempt to make the system legible to the adults who are, ultimately, responsible for their child's education. Several chapters are addressed directly to parents who are sceptical of AI in education, and they do not argue those parents out of their scepticism — they give them enough information to form a more precise opinion. If you are weighing up AI tutoring against a human tutor, the book does not tell you which to choose; it tells you what each thing actually is.
The bilingual decision reflects a simple reality. A significant proportion of KS3 families in the UK speak Mandarin at home, and many parents in those families are highly educated, deeply engaged in their child's schooling, and underserved by educational resources that assume English-only households. Publishing 《英国中学的AI私教》 alongside the English edition is not a marketing move; it is an acknowledgement that the book's audience exists on both sides of that language boundary.
What the book gives away — and what it does not
Some products guard their methodology. This one does the opposite. The four-level hint ladder, the weekly energy system, the safeguarding hard rules, the reasoning behind each tutor's personality — all of it is in the book, explained clearly enough that a technically inclined reader could build something similar.
That choice is intentional. Harewood's view, documented in the founder story, is that the value in an AI tutoring service is not the secrecy of its method but the quality of its execution. Parents who understand what a good AI tutor does are better placed to use it well, to spot when it is not working, and to make an informed decision about whether it is right for their child.
The companion service at aitutors.me is where the method runs in practice. The book is where the method is explained. They are designed to be read alongside each other, not as competing products.
Who should read it
The book is written for parents of KS3 children — ages 11 to 14, roughly Years 7 to 9 — who want more than a setup guide. It assumes no technical background and no particular familiarity with how AI works. Part II in particular starts from first principles: what is a language model, what is a token, why does "personality" emerge from training, and why does any of this matter for a child doing geography homework.
Curious Year 9 students will find the book accessible too; several chapters are written in a voice that addresses the student directly. The educators' appendix in Part VI is aimed at teachers who want to understand what their students are using and how it relates to curriculum and assessment objectives.
AI Tutors for Key Stage 3 and its Simplified Chinese edition are coming to Kindle. For availability and early access, visit aitutors.me/books.