“Show me your paper”
Ask to see the working, not just the answer. Scribbles and crossings-out are a great sign — they mean real thinking happened.
Our approach
Learning starts with your own thinking: a sketch, a question, a first attempt. We work through ideas together on paper and at the whiteboard, then use technology when it helps you see something more clearly. You learn how to work independently and how to use AI thoughtfully.
My lessons make room for hands-on thinking: pencil, paper, whiteboard, books, and conversation.
My attempt: w(w + 3) = 40
w² + 3w + 40 − 40 = 0
StudentHere's my working. Don't fix it — just tell me which line has the mistake.
AI study partnerLine 2: check the sign when 40 moves across the equals sign. Want a hint about why?
Why pencil first
This isn't nostalgia. Physical materials do three things for learning that screens — and especially AI — tend to undo.
A crossed-out line shows exactly where the thinking went wrong — so we can fix the reasoning, not just the answer. A deleted line or a regenerated response teaches nothing.
Writing by hand forces you to decide what matters before you write it down. That small act of choosing is where understanding gets built.
In-class tests, timed exams, oral presentations, interviews: the moments that count still happen without AI. Students should walk into them already fluent.
In vivo / live understanding
Every assignment we work on includes an in vivo check: students answer questions in person or on a call, without seeing the questions in advance. They explain their answers, show their reasoning, and apply the idea to a fresh question.
The student is responsible for understanding what they submit. These live checks help prevent cheating and reveal where the learning needs more attention. I’ve used this approach with college students to great effect.
The student teaches the idea to us, out loud, with no notes. Gaps show up in seconds.
Understanding can't be pasted.
A fresh problem the student has never seen, solved live by hand, thinking aloud.
There's no time to look it up — only to reason.
We tweak a problem the student just solved. Memorised answers collapse; real understanding adapts.
The variation is invented on the spot.
A paragraph or outline written by hand in the session, then marked up and rewritten together.
We see every word arrive.
Before running code or checking a graph, the student commits to a prediction on paper.
The prediction is theirs, right or wrong.
Find the mistake in a worked solution — sometimes one an AI produced. Spotting errors is a higher skill than avoiding them.
It trains the judgement AI lacks.
AI as a tool
Banning AI doesn't work, and pretending it doesn't exist leaves students to work it out alone. So we teach it like any other powerful tool: when to pick it up, how to use it well, and when to put it down.
Used this way, AI becomes a patient explainer, an endless source of practice questions, and an honest critic — available at 11 pm the night before a test.
AI comes after an honest try on paper — never before it. The struggle is where the learning is.
“Explain this another way”, “give me a hint”, “quiz me” — not “solve this” or “write this for me”.
AI is fluent and often wrong. Checking its claims, math, and code is a skill we practise on purpose.
If you can't explain it without the AI open, it doesn't go in your work.
Your school's and teacher's AI policies come first. When AI helped, say so.
We never use AI to help students cheat or to complete assignments for them. We teach AI as a holistic learning partner — to explain, to quiz, to critique work the student wrote — always toward the shared goal of real learning, and always within the school's rules.
My experience building software informs how I teach with technology: question the output, check the reasoning, and understand the work yourself. Meet your tutor
Try it
Eight real situations. For each one, decide: does this use of AI build learning, or replace it? Good for students — and surprisingly good for parents too.
Asking AI to explain a textbook step three different ways until one clicks.
Tool. The student is still doing the understanding — AI is just offering more angles in.
Pasting the essay prompt into a chatbot and handing in what comes back, lightly edited.
Crutch. All of the thinking — the actual assignment — was done by the machine. (And it usually breaks school rules.)
Solving a problem on paper, then asking AI to point out the line with the mistake — without fixing it.
Tool. Attempt first, targeted feedback second, correction by the student. That's how good tutoring works too.
Having AI write ten practice problems like Friday's quiz, with the answers hidden.
Tool. More practice is almost always good — as long as the student does it on paper and checks the answers carefully.
Copying a working function from an AI assistant into a project without being able to explain it.
Crutch. The code runs, but the student can't debug, extend, or defend it. Rule four: own every line.
Reading an AI summary of the chapter instead of the chapter itself.
Crutch. Summaries strip out the evidence and voice you'll need for analysis. (Reading first, then comparing your own summary with AI's? That's a tool.)
Asking AI to argue against your thesis so you can make your essay stronger.
Tool. AI as sparring partner: the student still has to answer the objections in their own words.
Checking every homework answer with AI before attempting the problem.
Crutch. Same tool, wrong order. Knowing the answer first removes the productive struggle that makes it stick.
Turn on JavaScript to answer — the explanations are shown below each situation.
Growing with the student
AI skills are introduced gradually, always after the fundamentals — and always with a parent's account and consent where a platform's age rules require it.
Reading aloud, counting, drawing and writing by hand. We build the habits every later tool depends on: try first, check your work, explain it out loud.
Nearly everything on paper. AI appears only together with the tutor — as a curiosity, and as a way to practise catching its mistakes.
Self-quizzing, alternative explanations, vocabulary help. Graded work stays pencil-first, and the student can explain everything in it.
Feedback on drafts, counter-arguments, test cases for code. Timed exam practice itself stays strictly AI-free — on paper, against the clock — because the exam will be.
Pair-programming, research help with careful verification, and the disclosure norms of each course and field.
For parents
Ask to see the working, not just the answer. Scribbles and crossings-out are a great sign — they mean real thinking happened.
If AI helped with homework, ask the student to explain the result in their own words. If they can't, the learning isn't finished yet.
A simple household rule: ten honest minutes on paper before any app, calculator, or chatbot comes out.
Where it's taught
Worked by hand, step by step — then checked, never outsourced.
Close reading and handwritten drafts, marked up and rewritten together.
Trace it on paper, then code it — up to a specialised AI & ML track.
Next step
Tell us the subject and what's been hard. We'll reply within one business day, and the intro call is free.