T-Twice Think Twice · AI Math Tutor

The complete
engagement system

Three dimensions of a single principle: that every student — even the least motivated — genuinely builds their reasoning, that teachers feel supported not watched, and that typing maths never blocks a thought again.

01
The founding conviction
No student is mediocre. Every student has their own pace, their own way of thinking — and every student can find pleasure in mathematics, if the experience is designed around them rather than against them.
The real problem was never motivation. It was the absence of a system that meets students exactly where they are — corrects their errors with respect, lets them progress at their rhythm, and reveals their own patterns of thought back to them. T-Twice doesn't force the student to change. It shows them who they already are — and who they're becoming. Without them even realising they're applying the cognitive mechanisms that actually work.
❌ What the student must NOT feel

I'm learning. I'm being evaluated. I'm working. I have to.

✓ What the student must feel

I'm discovering something about myself. I want to know if I was right. I'm getting somewhere — at my own pace, in my own way.

Phase 1
Entry
Create the tension before the effort
1
Intelligent timing
The student is always free to open T-Twice whenever they choose. But the AI observes their concentration patterns and natural availability — and suggests the right moment, never demands it. Even the most reluctant student gets an entry point they can say yes to.
"Got 10 minutes? One question about what blocked you yesterday."
2
The prediction hook — curiosity before effort
Before any working out, a 5-second intuition question. The student commits to a gut answer. Their brain is now invested — they need to know if they were right. They cannot leave without finding out. This works especially well for students who feel disconnected from maths: it makes it personal before it becomes technical.
"Does this sequence converge? Instinct only. 5 seconds."
"→ You said yes. 68% of students say the same. Now build your proof."
Phase 2
Engagement
Sustain without forcing
3
Targeted Socratic guidance
The student writes their reasoning. T-Twice identifies the exact error among 13 research-backed error types, then asks ONE question — targeted at that specific reasoning, in that specific course, at that specific stage. The professor's course is the only reference. The student's own thinking is always the starting point.
"Good idea. But you're applying Cauchy's criterion without checking whether the space is complete. What space are we in here?"
4
The almost-right — productive friction
Never "wrong" — always "80% there". The brain cannot tolerate being almost correct. That positive frustration becomes the engine of progress. For a student who's used to simply failing, this reframe is transformative: they're not bad at maths, they had one key thing to fix.
❌ Never
"Wrong. Try again."
✓ Always
"Your reasoning is 80% right. One key error at the limit step."
5
The hidden pattern — self-recognition
T-Twice detects patterns the student doesn't know about themselves yet. This moment of self-recognition is the most engaging in the entire system — it's personal, not academic. It creates an emotional bond with the tool that no reward or gamification system can replicate. And it respects each student's unique way of thinking.
"You only make this error when there's a square root in the problem."
"You succeed when you go back to the definition. Never when you skip straight to the result."
"Your reasoning gets shorter in the evenings — and less rigorous."
Phase 3
Revelation
The dopamine moment
6
Earned validation + gentle comparison
When the reasoning is right, T-Twice brings back the initial intuition and shows the gap. The confrontation between gut feeling and built logic is what stays in memory. The anonymous peer comparison creates aspiration without humiliation — crucial for students who have internalised a negative self-image around maths.
"You said it converged. You were right — but the harmonic divergence almost caught you."
"Logical gaps this week: 3. First-year average at this stage: 8. You're moving faster than most."
7
Active metacognition — becoming your own observer
At the end of the session — never after each question. T-Twice turns the question back to the student. They build their own diagnosis. This is the transition from guided learner to autonomous thinker. The student isn't just correcting errors anymore — they're understanding how their own mind works. Without knowing it, they're applying spaced retrieval, error-driven learning, and self-regulated cognition. The research mechanisms that work — invisibly, naturally, at their own pace.
"Why do you think you made that error?"
"You're hesitating less than 3 days ago on this type of question. Have you noticed?"
"What would you change in your reasoning if you started again?"
Phase 4
Retention
Make them want to come back tomorrow
8
Binge effect + identity + real stakes — three levers, one moment
At the end of the session, T-Twice activates three mechanics simultaneously: a door left open to the next challenge, a visible identity narrative around who the student is becoming mathematically, and a concrete link to the real exam ahead. For a student who used to avoid maths, these three things change what maths means to them.
⚡ Binge effect
Just one more
"A slightly harder variant — just to confirm what you just proved to yourself."
🪞 Identity
You're changing
"Strong intuition, rigour catching up — it shows in your data."
📅 Real stakes
Exam in 35 days
"At this pace: 1 logical gap in the exam instead of 6."
The complete optimal formula
Smart timing + Prediction hook + Socratic guidance + Almost-right + Hidden pattern + Earned validation + Metacognition + One more + Real stakes Autonomy
The student doesn't feel like they're learning. They feel like they're discovering who they are. The pedagogy is unchanged. The lived experience is transformed. That's the difference between a good educational tool and something genuinely addictive — even for those who never thought maths was for them.
Weeks 1–2
Curiosity enters
The student comes because they want to know if they were right. The prediction hook does its work — even for the most reluctant student.
Weeks 3–6
Identity anchors
They see themselves as someone who is progressing. They come back to confirm who they're becoming — at their own pace, in their own way of thinking.
Weeks 7–12
Autonomy arrives
They no longer need the tool. That is the system working exactly as designed — the student has internalised the cognitive habits that make learning stick.
02
1
Always propose, never instruct — "if you'd like", "you'll know better than us"
2
Acknowledge the teacher's work before any observation — they are the expert, T-Twice is the assistant
3
Speak about students with care — no negative labels, always a constructive reading
4
Short, precise messages — one concept, one observation, one open question. Never more than two lines.
01
30 seconds
Upload
An existing course PDF — nothing to create from scratch
02
Automatic
Extraction
T-Twice reads definitions, notations and course progression
03
2 minutes
Pedagogical profile
5 simple questions about how they teach — done once, remembered always
04
Automatic
Calendar
Built automatically via assignments — no manual input
1
The suggestion of the day
A short, precise observation about the class, rooted in a specific concept. An open question at the end — the teacher decides alone whether it's relevant.
"Uniform continuity — several students are using the property as if it were the definition. Worth revisiting?"
2
The caring watch
T-Twice quietly notices students who seem to be going through a difficult period. Not to label them — to allow the teacher to give them a thoughtful look if they see fit.
"Léa and Thomas have been stuck on Cauchy sequences since Monday — they did well in the previous chapter. You'll see them in tutorial on Friday?"
3
The pedagogical mirror
When something is working well in the class, T-Twice notices and says so — simply, without exaggerating. The teacher deserves to know their investment is paying off.
"Since your Tuesday tutorial on proof writing, logical gaps have fallen 40% across the class. Something worked really well."
4
The class portrait
A reading of the collective dynamic, updated each week. Not to categorise — to help the teacher adapt their approach if they judge it useful.
"Your class grasps ideas quickly but still takes time to formalise them — often the sign of curious students gaining in rigour."
5
Informed preparation
Before a tutorial or lecture, T-Twice shares what it has observed on upcoming concepts. Not to dictate preparation — to enrich the teacher's thinking if they choose.
"For your next tutorial on Cauchy sequences — the confusion between Cauchy sequence and convergent sequence comes up often. You'll know what to do with it."
6
Error mapping
T-Twice aggregates errors across the class and presents them clearly. The teacher sees collective patterns that would be invisible exercise by exercise.
"Uniform continuity — the definition/property confusion appears in several papers this week. Perhaps a collective reminder, when you get the chance."
7
What actually helps these students
T-Twice shares what it observes about remediation approaches that seem to help — not as a truth, but as a lead. The teacher remains the judge of what suits their group.
"For this type of continuity error — students seem to progress better with a counterexample than with a new definition. Maybe something to try, if it feels right."
Pedagogical questionnaire — 5 questions, once only
"In your course, what comes first — the rigorous definition or the intuition?"
"When a student makes an error, do you prefer to ask a question, explain directly, or show a counterexample?"
"Are you most attentive to those who drift in silence, collective progress, or those who move ahead fast?"
Weeks 1–2
Discovery
The teacher sees what they couldn't see alone about their class
Weeks 3–5
Adaptation
They adapt their teaching with greater confidence and serenity
Weeks 6–8
Visible impact
They see their students progressing differently
Week 9+
Natural integration
T-Twice becomes part of how they teach — a quiet voice in the background
03
❌ Without an adapted keyboard

30 seconds searching for a symbol. The thread of reasoning is broken before it even began. Technical friction kills mathematical thought — especially for students already low on confidence.

✓ With the T-Twice keyboard

Typing stops being an obstacle. The student thinks about their reasoning — the keyboard handles the rest. Within 3 weeks, it's a reflex.

1
Natural language
Write the way you speak
Type in plain language, the keyboard converts in real time. Reversible word by word.
for all epsilon → ∀ε
limit of u_n → lim uₙ
2
Contextual keyboard
The right symbols for the exercise
Adapts to the chapter AND the problem statement. Works on all devices: phone, tablet, floating desktop bar.
Cauchy → ε δ n₀
Algebra → ker det λ
Probability → P(·) Ω E[X]
3
Text shortcuts
A reflex by session 3
Like phone autocorrect. After a few sessions, faster than any other keyboard — and fully personalised.
-> → · epsilon → ε
infty → ∞ · n0 → n₀
4
Voice input
Dictate out loud
T-Twice transcribes and converts simultaneously. Ideal on phone. Voice frees the hands to think.
"for all epsilon" → ∀ε
5
Notation guidance
Form, never content
Suggestions of frequent notations for the current chapter. The student chooses or ignores. The reasoning is never touched.
Cauchy: ∀ε>0 ∃n₀ · R complete
6
Full control
The student decides everything
Toggle to disable. Word-by-word reversal for any conversion — text, shortcut or voice. No surprises, ever.
7
Progressive learning
The keyboard learns the student
Observes habits, surfaces favourite shortcuts, personalises session after session — respecting each student's natural rhythm.
S1: searching · S3: reflexes
S6: thinking about reasoning
8
Multimodal
Same experience everywhere
Same shortcuts, same logic across all devices. One learning curve. Voice everywhere — the only constant in a student's unpredictable day.
📱 Phone
Keyboard + voice
On the bus, in motion. Voice input takes over when hands are needed elsewhere.
📟 Tablet
Handwriting + voice
Handwritten mathematical notation recognition. Natural for complex problem-solving.
💻 Computer
Floating bar + voice
Late at night, working on a problem set. The floating bar is always there. No interruption to the flow.
Week 1
Discovery
Shortcuts, voice, contextual keyboard. The student explores what's available.
Week 2
Reflexes
Shortcuts become automatic. Typing starts to disappear from awareness.
Week 3+
Total fluency
No keyboard — only reasoning. The input obstacle is gone for good.

Three pillars.
One principle.

That every student — especially the ones who gave up — builds their reasoning with pleasure, at their own pace, in their own way. That teaching is supported, not surveilled. That the technology disappears behind the thinking.

Student Pillar
Progress every student — including the one who stopped believing
  • 4-phase session: curiosity → engagement → revelation → retention
  • Prediction hook creates investment before effort
  • Targeted Socratic guidance, one question at a time
  • 13 research-backed error types, precisely identified
  • Hidden pattern recognition — self-knowledge through maths
  • Active metacognition — autonomous learning without noticing it
  • Long-term arc: curiosity → identity → autonomy
Teacher Pillar
Supported, not replaced
  • Onboarding under 2 minutes, once only
  • Daily suggestion — a lead, never an instruction
  • Pedagogical mirror showing real impact
  • Weekly class portrait for collective insight
  • Error mapping across the entire group
  • Caring watch for silent dropouts
Keyboard Pillar
Type nothing. Think everything.
  • Natural language → symbols in real time
  • Contextual keyboard by chapter and problem
  • Automated text shortcuts — reflex by session 3
  • Multiplatform voice input
  • Full control — word-by-word reversal
  • Same experience across phone, tablet, computer
The result — T-Twice
Within 3 weeks, the student stops thinking about how to type their maths. Within 3 sessions, they stop looking for the answer. They start building their reasoning — at their own pace, in their own way, without ever realising they're applying the cognitive mechanisms that actually make learning last.