COURSE MASTERY
62%↗ +8%
Across 5 topics
studywise.
Your sources stay on this deviceAK
TUESDAY, OCTOBER 13, 2026
Good morning, Arjun ✦
A little progress today makes a big difference tomorrow.
YOUR NEXT BEST STEP
Revisit cellular respiration
Your practice shows this topic could use another pass. A 6-minute review is a good place to start.
Suggested from your last 2 practice sessionsSTUDY TIME
2.4 hrs
This week · +18% vs last weekQUESTIONS ANSWERED
12↗ +4
✓ 75% accuracy this week
Keep practicing to build recallPERSONALIZED FOR YOU
Topics to focus on
PICK UP WHERE YOU LEFT OFF
Recent activity
GROUNDED IN YOUR COURSE MATERIAL
Your AI tutor
Ask a question. Every answer links back to the exact source.
● Source grounded
✳
Studywise tutor Using 3 sources in this course
BUILD STRONGER RECALL
Practice & assessments
Create a quiz that adapts to what you know.
SET UP YOUR SESSION
What would you like to practice?
✦ Questions are created from your sources, verified against their answer key, and tracked to avoid repeats.
RETURN TO IT BEFORE IT FADES
Review & study plan
Short retrieval sessions, spaced over time, with course evidence one click away.
DUE TO REVIEW
0concept cards are readyNEXT EXAM
Set a datePlan your revision windowDAILY STUDY TIME
20 minyour chosen daily budgetRETRIEVAL PRACTICE
Ready when you are
Study plan
ⓘ
Why spaced review? Quizzing can strengthen later recall; useful review intervals depend on how long you need to retain the material. These starter intervals are a transparent heuristic—not a validated prediction of forgetting. Rate how well you recalled each idea, and the queue adapts.
ONE HOME FOR ALL YOUR COURSE MATERIAL
Course library
Upload sources. We’ll organize them into cited, searchable learning units.
YOUR MATERIAL
3 sources
How source-grounding works
Text is split into short, traceable passages. Each passage keeps its page, slide, or timestamp. Answers only use passages that match your question; gaps are called out clearly. Scanned page images and diagrams are detected as visual content but need OCR/caption text before they can be searched in this prototype.
HOW IDEAS CONNECT
Your learning path
Follow the prerequisites, then build toward connected understanding.
Strong foundation In progress Ready to learn
Learn in sequence
Start with the foundations at the top. Concepts below build on them, so strengthening prerequisites makes later topics easier.
CROSS-SOURCE CONSISTENCY
Check for conflicts
We compare how your lecture, slides, and textbook explain each topic and flag potential disagreements.
YOUR LEARNING PROFILE
Progress & mastery
A running estimate of what you know, updated after every answer.
OVERALL COURSE MASTERY
62%
Estimate based on 12 assessment and review responses.
Bayesian knowledge tracing · P(learn) 0.10 · guess/slip 0.20PERSONALIZED INSIGHT
You’re building a steady foundation.
Keep practicing your developing topics. Strong recall comes from returning to ideas over time.
TOPIC BY TOPIC
Your mastery estimates
MASTERY OVER TIME
SHOW YOUR WORK
Evidence behind your estimates
Mastery changes only after a learning check. Each event below links to the source used for that check.
LEARNING PATTERNS
Potential misconceptions
DIAGNOSIS HISTORY
What we've learned about your mistakes
MEASURE WHAT MATTERS
System evaluation
A transparent, reproducible benchmark for retrieval, grounding, and personalization.
ⓘ
Results are computed from the included course and benchmark questions. This is a small, local prototype benchmark—not a RAGAS/DeepEval external audit. Metrics are calculated live and labeled with their sample sizes.
KNOWN-ANSWER SET
Retrieval benchmark
SIMULATED LEARNERS
Personalization run
Synthetic profiles illustrate the update logic. They are not evidence of real student learning gains.
EVALUATION RUN HISTORY
Benchmark snapshots over time
Each benchmark run is saved locally so you can compare metrics as your course material changes.
TEAM TEST SET
Build your evaluation set
Add questions with expected source IDs (or mark as off-material). The judge scores faithfulness, answer relevancy, context precision and recall using your Gemini key.
METHOD & LIMITATIONS
Faithfulness proxy
Share of tutor response sentences that overlap a retrieved source passage by at least one informative token. It is a conservative lexical proxy, not a semantic judge.
Context precision / recall
Calculated against known source IDs in the fixture benchmark. With only six authored cases, results are directional.
Answer relevancy
Query-to-retrieved-context token overlap. This prototype has no external LLM judge or embedding model.
Repetition
Repeated question IDs divided by generated question count across simulated sessions.