What is Acadevo?
Acadevo is India's first truly adaptive learning platform purpose-built for students in classes 6 through 10. Unlike traditional ed-tech apps that serve the same static content to every student, Acadevo uses Item Response Theory (IRT), Bayesian Knowledge Tracing (BKT), and spaced repetition algorithms to build a unique learning path for each student in real time.
Every question a student answers informs the system about their ability level, and the next question is dynamically selected to sit in their Zone of Proximal Development, the sweet spot where learning happens fastest. The result is a practice experience that is never too easy, never too hard, and always moving students forward.
47,000+ Questions
A massive calibrated question bank spanning five subjects, 312 topics, and difficulty levels from foundational to expert.
Three Dashboards
Purpose-built interfaces for students, parents, and teachers with distinct tools, insights, and controls for each role.
Proven Science
Built on IRT, BKT, and SM-2 spaced repetition, the same frameworks used in GRE, GMAT, and leading global assessment systems.
Subject Coverage
| Subject | Topics | Questions | Classes |
|---|---|---|---|
| Mathematics | 72 | 8,640+ | 6-10 |
| Science | 72 | 7,200+ | 6-10 |
| Social Science | 72 | 5,760+ | 6-10 |
| English | 48 | 4,800+ | 6-10 |
| Hindi | 48 | 3,840+ | 6-10 |
| Total | 312 | 30,240+ | 6-10 |
Board coverage: All content is aligned to CBSE curriculum with ICSE mapping in progress. Questions are tagged by chapter, topic, and sub-topic for granular tracking.
The Adaptive Learning Engine
Powered by Item Response Theory (IRT)
Most learning apps assign questions randomly or in a fixed sequence. Acadevo's adaptive engine treats every practice session as a real-time assessment, continuously estimating each student's ability and selecting the optimal next question. This is the same family of algorithms that powers the GRE, GMAT, and other high-stakes adaptive tests, now available for daily practice.
Question Properties
Each question in the bank has three IRT parameters calibrated through data:
- a (Discrimination): How well the question differentiates between ability levels
- b (Difficulty): The ability level at which a student has a 50% chance of answering correctly
- c (Guessing): The probability of getting it right by chance (typically 0.25 for MCQ)
Student Ability (Theta)
Each student has a continuously updated ability estimate (theta) per topic, measured on a standard scale:
- Starts at 0.0 (average) for new students
- Ranges from -3.0 (foundational) to +3.0 (expert)
- Updated after every single response using maximum likelihood estimation
Selection Algorithm
The engine uses Fisher Information to select the next question:
- Calculates information value of every available question at student's current theta
- Selects the question with maximum information (most diagnostic value)
- Applies exposure control to prevent overuse of any single item
The Learning Zone
The engine keeps every student in their Zone of Proximal Development (ZPD), the range where questions are challenging enough to promote growth but achievable enough to maintain motivation.
Skill Levels
| Level | Theta Range | Description | Student sees |
|---|---|---|---|
| Foundational | -3.0 to -1.5 | Significant gaps in prerequisite concepts | Basic concept-building questions |
| Developing | -1.5 to -0.5 | Partial understanding, inconsistent application | Scaffolded practice problems |
| Proficient | -0.5 to +0.5 | Solid grasp of core concepts | Grade-level standard problems |
| Advanced | +0.5 to +1.5 | Strong command, can apply to novel situations | Application & analysis questions |
| Expert | +1.5 to +3.0 | Deep mastery, ready for competition-level work | Olympiad-style challenges |
BKT Mastery Tracking
Beyond ability estimation, the engine uses Bayesian Knowledge Tracing (BKT) to estimate the probability that a student has truly mastered each skill. BKT models four parameters: initial knowledge (P(L0)), probability of learning on each attempt (P(T)), probability of slipping (knowing but answering wrong, P(S)), and probability of guessing (not knowing but answering right, P(G)).
After each response, the system updates the mastery probability. A skill is considered mastered when P(mastery) exceeds 0.85. This ensures students are not just answering correctly by chance but have genuinely learned the concept.
For Students
A practice experience that adapts to you, every single session.
3.1 Daily Practice Targets
Students see a simple dashboard showing their daily practice goals across all subjects. Each subject requires 5 questions per day. As questions are answered, the rings fill up in real time, giving students a clear sense of progress and completion.
3.2 Practice Sessions
Each practice session follows a structured adaptive flow. The engine selects each question in real time based on the student's ongoing performance within the session.
Subject
optimal question
answers
updated
selected
summary
Sessions are 5 questions by default. After each session, students see a summary showing questions attempted, correct rate, and how their skill score changed. They can immediately start another session or switch subjects.
3.3 Skill Score
The Skill Score is a student-friendly representation of their theta value, normalized to a 0-100 scale. It provides a single, easy-to-understand number that summarizes how a student is performing in each subject.
3.4 Topic Mastery Map
Students can drill into each subject to see their mastery status across every topic. The mastery map provides a visual overview of where they stand and where they need to focus.
| Status | Indicator | Meaning | Example Topics |
|---|---|---|---|
| Mastered | P(mastery) ≥ 0.85 | Student has demonstrated reliable command of this topic | Fractions, Decimals, Basic Algebra |
| Developing | 0.50 ≤ P(mastery) < 0.85 | Making progress but needs more practice for consistency | Quadratic Equations, Geometry |
| Needs Work | 0.20 ≤ P(mastery) < 0.50 | Significant gaps identified, focused practice recommended | Trigonometry, Statistics |
| Not Started | No attempts yet | Topic has not been practiced | Probability, Mensuration |
3.5 Spaced Repetition Schedule
Once a student masters a topic, the spaced repetition system schedules periodic reviews at increasing intervals to ensure long-term retention. This is based on the SM-2 algorithm, the same science behind Anki and other memory systems.
3.6 Practice Regularity
A heatmap tracks daily practice consistency, similar to GitHub's contribution graph. Each square represents one day, and the color intensity shows how many questions were completed that day. Consistency is rewarded with streak tracking.
Darker squares = more questions practiced. Gray squares = no practice that day. Students and parents can see monthly and yearly views.
3.7 Free vs Premium
| Feature | Free Trial (7 days) | Free | Premium |
|---|---|---|---|
| Adaptive practice | ✓ All subjects | ✓ All subjects | ✓ All subjects |
| Questions per day | Unlimited | 5 per subject | Unlimited |
| AI Hints | ✓ Unlimited | ✕ | ✓ Unlimited |
| Skill Score | ✓ | ✓ | ✓ |
| Topic Mastery Map | ✓ | ✓ | ✓ |
| Spaced Repetition | ✓ | ✕ | ✓ |
| Parent Dashboard | ✓ | ✓ Basic | ✓ Full |
| Teacher Dashboard | ✓ | ✓ | ✓ |
| Practice Heatmap | ✓ | ✓ | ✓ |
| Detailed Analytics | ✓ | ✕ | ✓ |
| Price | Free for 7 days | Free forever | Rs 299/month |
For Parents
Complete visibility into your child's learning journey, updated in real time.
4.1 Linking to Your Child
Parents connect to their child's account using a unique linking code generated by the student. The process is simple and privacy-respecting, the parent never needs the student's password.
- Student opens their profile and taps "Generate Parent Code"
- A 8-character code is generated, valid for 24 hours
- Student shares the code with parent (verbally, text, etc.)
- Parent creates their account and enters the code
- Accounts are linked, parent sees real-time data
Sample parent linking code (expires in 24 hours)
4.2 What Parents Can See
Skill Score
See your child's current skill score across all subjects, with trend arrows showing improvement or decline over the past week.
Topic Mastery Map
Visual breakdown of mastered, developing, and weak topics in each subject. Identify exactly where your child needs help.
Practice Heatmap
See which days your child practiced and how much. Track consistency patterns and identify when engagement drops.
Year-over-Year Comparison
Compare performance across academic years. See how your child has grown since they started using Acadevo.
Daily Progress
Real-time view of today's practice: subjects attempted, questions answered, accuracy rate, and time spent.
Multiple Children
Link multiple children under one parent account. Switch between children seamlessly to monitor each one individually.
4.3 Academic Year Timeline
Acadevo preserves learning history across academic years, so parents can see the full arc of their child's educational growth.
2026-27 (Current)
Class 8 — Active since April 2026. Skill Score: 74. 1,240 questions completed.
2025-26
Class 7 — Completed. Final Skill Score: 68. 4,890 questions completed. 47 topics mastered.
2024-25
Class 6 — Completed. Final Skill Score: 55. 2,340 questions completed. 28 topics mastered.
4.4 Institution Visibility
If the student's school uses Acadevo, parents can see which class their child is enrolled in and who their teacher is. However, parents cannot see other students' data or class-wide analytics, maintaining privacy for all families. The teacher controls what aggregate insights are shared with parents via the institution settings.
For Teachers
Powerful tools to monitor, understand, and support every student in your class.
5.1 Setting Up a Class
Teachers can create classes in under a minute. Students join using a simple invite code.
Account
Class
Invite Code
Join
Dashboard
Sample class invite code (Math, Class 8, Section A)
5.2 Class Dashboard
The class dashboard gives teachers a bird's-eye view of every student's performance, with the ability to drill into individual detail.
| Student | Skill Score | Level | Mastered | Developing | Needs Work | Last Active |
|---|---|---|---|---|---|---|
| Aryan Sharma | 82 | Advanced | 14 | 6 | 2 | Today |
| Priya Patel | 74 | Proficient | 10 | 9 | 3 | Today |
| Rahul Gupta | 58 | Developing | 5 | 8 | 9 | Yesterday |
| Sneha Reddy | 41 | Foundational | 2 | 5 | 15 | 3 days ago |
Skill Distribution
Visual breakdown of where your class stands:
1 Foundational, 1 Developing, 1 Proficient, 1 Advanced, 0 Expert
Class average: 63.8 (Proficient)
5.3 Student Detail View
Clicking on any student opens a detailed view showing their complete learning profile: skill score trend over time, topic-by-topic mastery breakdown, practice frequency, spaced repetition adherence, and AI hint usage patterns. Teachers can use this to plan targeted interventions, identify students who need extra support, and recognize high performers.
5.4 Multiple Classes
Teachers can create and manage multiple classes simultaneously, one for each subject-section combination they teach. Each class has its own invite code, student roster, and analytics. Switching between classes is instant from the teacher dashboard, and aggregate views can show performance across all classes at a glance.
5.5 Institution Management
School administrators can create an institutional account that links all teachers and classes under one umbrella. This enables school-wide analytics including aggregate skill score distributions across grades, subject-wise performance comparisons, teacher engagement metrics, and identification of at-risk students. Administrators can also manage teacher access, set default configurations, and export compliance reports.
AI-Powered Explanations
When students get stuck, they don't just see the right answer. They get a personalized, step-by-step explanation generated by Claude AI that teaches them how to think through the problem.
Step 1: Student Gets Stuck
After answering a question wrong (or tapping "Need a hint"), the student can request an AI explanation. The system captures the question, the student's wrong answer, and their current skill level.
Step 2: AI Generates Explanation
Claude receives the question context, the student's ability level, and pedagogical guidelines. It generates an explanation at the right level of complexity for that specific student.
Step 3: Student Learns
The explanation is displayed with clear formatting, step-by-step reasoning, and a final summary. The student can then re-attempt a similar question to reinforce the concept.
Real Example
Question: A train travels from City A to City B at 60 km/h and the distance is 960 km. How long does the journey take?
Student's answer: 960 km/h (confused distance with speed)
AI Hint:
I see you wrote 960 km/h as the answer. Let's think about what we know and what we need to find.
Given: Distance = 960 km, Speed = 60 km/h
Find: Time = ?
Remember the formula: Time = Distance / Speed
So: Time = 960 km / 60 km/h = 16 hours
Tip: Always check your units! The question asks for time (hours), not speed (km/h). Distance is what was given (960 km), not the answer.
Caching System
To keep costs low and response times fast, Acadevo uses an intelligent caching system for AI hints.
Claude Integration: Acadevo uses Anthropic's Claude API for hint generation. Each hint request includes the question text, answer options, correct answer, student's response, and the student's ability level. Claude is instructed to generate age-appropriate, curriculum-aligned explanations that teach reasoning rather than just providing the answer. Responses are cached by question ID for reuse across students.
Remember What You Learn
Without review, students forget up to 80% of what they learn within a month. Acadevo's spaced repetition system fights the forgetting curve by scheduling intelligent reviews at precisely the right time.
The Forgetting Curve
SM-2 Algorithm
Acadevo uses a modified SM-2 algorithm (the same core algorithm used by Anki) to determine optimal review intervals. The algorithm adjusts intervals based on how well the student performs during each review. Correct answers increase the interval; incorrect answers reset it to a shorter duration.
| Review | If Correct | If Wrong | Note |
|---|---|---|---|
| 1st correct review | 1 day | 3 days (retry) | Immediate consolidation |
| 2nd correct review | 6 days | 3 days (retry) | Short-term retention check |
| 3rd correct review | 15 days | 3 days (retry) | Medium-term retention |
| 4th correct review | 38 days | 3 days (retry) | Long-term retention building |
| 5th correct review | 94 days | 3 days (retry) | Deep long-term memory |
How Students Experience It
Spaced repetition is woven seamlessly into the daily practice flow. When a student opens a subject, the engine checks if any mastered topics are due for review. If so, 1-2 review questions are mixed into the regular practice session. Students see a small "Review" badge on these questions so they know it's a retention check rather than new learning. This means students don't need to do anything special; the system handles the scheduling automatically.
Analytics & Insights
Acadevo provides layered analytics for every stakeholder: administrators see platform health, teachers see class performance, and parents see individual progress.
Admin Analytics
Platform administrators have access to a comprehensive analytics dashboard covering:
- Daily Active Users (DAU): Real-time count of students who practiced today
- Registrations: New sign-ups by day, week, and month with source attribution
- Sessions: Average session length, questions per session, sessions per day
- Conversion Funnel: Tracking from registration through activation to paid conversion
- Revenue: MRR, subscription upgrades, churn rate, LTV metrics
- Content Health: Question bank utilization, item difficulty drift, flagged questions
Conversion Funnel
Teacher Class Analytics
Beyond the class dashboard table, teachers get detailed analytics including: class-wide skill score distribution over time, topic-level heatmaps showing which topics the class struggles with most, engagement metrics (active vs. inactive students, average practice frequency), and exportable reports for parent-teacher meetings. Teachers can identify intervention targets, such as a topic where more than 40% of the class scores below "Developing," or a student whose engagement has dropped in the last two weeks.
Content Health Monitoring
The platform continuously monitors question bank health through automated statistical checks. Questions with abnormal discrimination parameters (a < 0.3, indicating the question doesn't differentiate well between ability levels), extreme difficulty values (|b| > 3.0), or high guessing parameters (c > 0.4) are flagged for human review. The system also tracks exposure rates to ensure no single question is overused, and identifies topics where the question bank may need expansion based on student attempt patterns.
Product Roadmap
Where we are, where we're going.
Q2 2026 — Foundation Launch
12 features shipped:
- Adaptive practice engine (3PL IRT)
- Student dashboard with skill scores
- Topic mastery map with BKT
- Daily practice targets (5 subjects)
- Parent dashboard with linking
- Teacher dashboard with class analytics
- AI-powered hints (Claude integration)
- Hint caching system
- Spaced repetition (SM-2)
- Practice heatmap
- Razorpay payment integration
- 30,240+ calibrated question bank
Q3 2026 — Engagement & Growth
6 features in development:
- ICSE board curriculum mapping
- Gamification layer (badges, streaks, leaderboards)
- Peer comparison (anonymized percentile)
- Offline mode for low-connectivity areas
- WhatsApp parent notifications
- Expanded question bank (+17,000 items)
Q4 2026 — Depth & Intelligence
5 features planned:
- AI-generated practice tests (chapter-wise, full-length)
- Prerequisite skill graphs (automatic remediation)
- Voice-based hints in Hindi/English
- School admin dashboard with multi-class views
- API for third-party LMS integration
2027 — Scale & Expansion
6 features planned:
- Classes 1-5 curriculum (primary school)
- Regional language support (Tamil, Telugu, Kannada, Marathi)
- Competitive exam prep (NTSE, Olympiads)
- AI tutor (conversational learning mode)
- Parent mobile app (native iOS/Android)
- International curriculum support (Cambridge, IB)
Technical Specifications
The technology stack and standards that power Acadevo.
Architecture
- React 18 (frontend SPA)
- Node.js + Express (backend API)
- Cloud Firestore (primary database)
- Firebase Authentication (auth provider)
- Claude API by Anthropic (AI hints)
- Razorpay (payment processing)
- Region: asia-south1 (Mumbai)
Adaptive Engine
- 3-Parameter Logistic IRT (3PL)
- Bayesian Knowledge Tracing (BKT)
- SM-2 Spaced Repetition Algorithm
- Fisher Information item selection
- Maximum Likelihood Estimation (MLE)
- Exposure control via Sympson-Hetter
- Real-time theta updates per response
Data & Privacy
- All data stored in India (asia-south1)
- TLS 1.3 encryption in transit
- AES-256 encryption at rest
- No third-party data sharing
- GDPR-aligned data practices
- Parent consent required for minors
- Data export available on request
Availability & Performance
- 99.95% uptime SLA
- Global CDN (Firebase Hosting)
- Average API response: <200ms
- Question selection: <50ms
- AI hint generation: <3s (uncached)
- AI hint cached: <100ms
- Auto-scaling backend infrastructure