Skip to content
Features

Everything a real tutor does.
Just at the speed of software.

An overview of the systems powering TutorMate - adaptive learning, the seven modes, study tools, test prep, RAG, and the accessibility layers underneath it all.

Adaptive engine

Misconception detection, mastery scoring, frustration sensing and pace calibration on every turn.

Learn more →

11 AI buddies

A different teaching personality for every subject — from Socratic math coaching to playful language roleplay.

Learn more →

7 interaction modes

Socratic, step-by-step, roleplay, examples, quiz, show-your-thinking, and co-learn.

Learn more →

Study tools

Turn any PDF, video or lecture into flashcards, mind maps, audio lessons and study guides.

Learn more →

Test prep

AP, SAT, ACT, IELTS, TOEFL, GRE, LSAT and NY Regents — rubric-locked, score-band predictions.

Learn more →

RAG & ingestion

PDFs, slides, audio, video and handwritten notes become a searchable, cited knowledge base.

Learn more →

Age-adaptive UI

The same product reshapes itself across K–12 — typography, density and tone all shift by grade band.

Learn more →

Accessibility

High-contrast, dyslexia-friendly fonts, reduced motion and a screen-reader-first transcript — WCAG 2.1 AA.

Learn more →
01 · ADAPTIVE ENGINE

The tutor watches every signal — and changes course.

Four overlapping layers run on every interaction: misconception detection, mastery scoring, frustration sensing, and pace calibration. The session you start at 4pm is different from the one you'd have had at 9am.

  • Real-time misconception diagnosis — not just right/wrong
  • Bayesian mastery scoring per topic, per student
  • Frustration detection from hesitation, hint use, retries
  • Pace recalibration every 3–5 turns
SIGNALS · LAST 60S
PaceSLOWING
Hint depth usedTIER 2 OF 3
ConfidenceRECOVERING
Topic mastery0.71 → 0.73
DECISION · Switch to visual mode. Offer worked example before next problem.
02 · HINT PROGRESSION

Three hints before the answer. Server-enforced.

We borrowed the layered hint model from Khanmigo and made it stricter: the model never has the answer to give. Three hint tiers are surfaced in order, with a worked example as last resort — never the answer outright.

  • Tier 1: directional nudge — "what are you noticing?"
  • Tier 2: focusing question — "what does the denominator do?"
  • Tier 3: partial scaffold — first step shown, rest hidden
  • Final: worked example with the same shape, different numbers
PROBLEM · 1/2 + 1/3 = ?
1
TIER 1 · NUDGE
"What do you notice about the two denominators? Are they the same size of slice?"
2
TIER 2 · FOCUS
Locked — used after Tier 1 has been considered.
3
TIER 3 · SCAFFOLD
Locked.
WORKED EXAMPLE
Locked — different problem, same pattern.
03 · STUDY TOOLS

Any source, every format.

Paste a URL, drop a PDF, upload an audio lecture, link a YouTube video. We extract the structure, build a topic map and let you study it however works — flashcards, mind maps, audio lessons, personalized notes.

  • Flashcards with Leitner scheduling — exportable to Anki
  • Mind maps that grow as you study
  • Audio lessons narrated in your buddy's voice
  • Personalized study guides tuned to your weak spots
  • OCR for handwritten homework via Tesseract
SOURCE → ARTIFACTS
Chapter_7_Photosynthesis.pdf
2.1 MB · 18 PAGES
EXTRACTED
FLASHCARDS
42
MIND MAP NODES
18
AUDIO
14:32
STUDY GUIDE
3 pg
04 · RAG & INGESTION

The tutor reads what you give it.

A first-class document ingestion pipeline: semantic chunking, vector embeddings via pgvector, and cosine retrieval — so the tutor cites your textbook, not the open web.

  • PDF, DOCX, PPTX, EPUB, plain text, code files
  • URL scraping with article extraction
  • YouTube + audio transcription via Whisper
  • Image OCR for whiteboards, worksheets, handwritten notes
  • Cited responses with source line numbers
INGESTION PIPELINE
Document received
PDF · 18 pages · 4,212 tokens
Semantic chunking
42 chunks · 256 tokens avg
Embeddings · text-embedding-3
1,536 dim · 42 vectors
Indexed → pgvector
READY · 38ms search
05 · TEST PREP

Practice that maps to scores.

Rubric-locked grading using the same descriptors the real exams use. Predicted scores that move with you. AP tracks aligned to the official Course and Exam Description for every subject.

  • SAT — adaptive section practice, ±20 band prediction
  • ACT — composite tracking with ±1 accuracy
  • AP — 38 subjects, FRQ rubric-graded
  • IELTS — descriptor-locked, voice speaking practice
AP Exams20+ subjects
Regents ExamsMath · Science · ELA · SS
SATReading & Writing · Math
ACTEnglish · Math · Reading · Science
IELTSAcademic & General
TOEFL iBTUniversity admissions
GREVerbal · Quant · AWA
LSATLogical reasoning · RC
SAT · PRACTICE 04
1,420
PREDICTED COMPOSITE · ± 20
Math720 → 750
Reading & Writing650 → 670
Inference questions in the Reading section are still the biggest score lift available. 18 untouched practice items remaining.
06 · STANDARDS

Every topic maps to a standard.

Common Core, NGSS, state standards — each lesson, assessment and mastery score is tagged to its standards code. Defensible reporting for IEPs, parent meetings and district reviews.

  • Common Core (Math & ELA), NGSS (Science)
  • All 50 state standards
  • Cambridge, IB, AP Course & Exam Descriptions
  • Custom district scope & sequence import
CURRENT TOPIC · 8.EE.A.1
Know and apply the properties of integer exponents to generate equivalent numerical expressions.
8.EE.A.18.EE.A.28.EE.A.38.EE.A.4
PROGRESS TOWARD MASTERY
Prereq: 6.EE.A.1
Current: 8.EE.A.1
Next: 8.EE.A.2
07 · AGE-ADAPTIVE UI

The same product, four ages.

Font sizes, touch targets, corner radii, animation duration and vocabulary all shift by grade band. The Grade 11 UI is dense and crisp. The Kindergarten UI is gentle, oversized and forgiving.

K – 2

Gentle & large

56px touch targets, 1.25rem radii, 1.03rem base text, 480ms animations.

G3 – 5

Friendly

1.1× spacing, 1rem radii. Buddy reactions front-and-center.

G6 – 8

Standard

System default. Where the design system was tuned.

G9 – 12

Crisp & dense

0.95× spacing, 0.5rem radii, 240ms animations. Power-user mode.

08 · ACCESSIBILITY

Built for every learner.

Four adaptation layers ship by default: high-contrast mode, dyslexia-friendly font, reduced motion, and grade-band theming. WCAG 2.1 AA throughout — focus rings, ARIA landmarks, keyboard nav.

  • High-contrast: pure white background, no tints
  • Dyslexia font: OpenDyslexic + Atkinson Hyperlegible fallback
  • Reduced motion: all animations collapse to 0.001ms
  • Screen-reader-first chat transcript with role tags
  • Voice-only mode for visual fatigue or low-vision use
ACCESSIBILITY MENU

See it on real students.

Free for one teacher and one classroom. Pilot in a week.

Start freeFor teachers