Full-stack learning platform that builds personalized study plans from a student's notes and exam dates, diagnoses exactly why they are getting questions wrong across six failure modes, and serves targeted practice each day using an ever-growing concept knowledge map.
Live at veridian.study · Currently in closed beta
Most students don't fail because they aren't capable. They fail because they study the wrong things, study passively, and have no real feedback loop between effort and outcome until a grade comes back. By then, it's too late to change anything.
Every one of these failures is solvable with known learning science:
- Spaced repetition solves the timing problem
- Active recall solves the method problem
- Source-grounded AI question generation solves the material alignment problem
- Knowledge gap detection solves the feedback problem
Veridian implements all four, integrated into a single system that requires no prior knowledge of how to study effectively.
A Journey is the core unit — one per subject or exam. You create one by uploading your source material: class notes, a textbook chapter, a PDF, or raw pasted text. Veridian extracts the knowledge structure inside and builds a personalized study plan calibrated to your exam deadline.
Inside each Journey are Modules — individual concept units extracted automatically from your source material. Each Module has its own mastery state, FSRS review schedule, and stage placement tracked independently.
Every Module moves through three stages. Stage placement is determined by the diagnostic — not a fixed schedule.
Stage A — Learn
For first exposure. A structured Learning Guide breaks the concept into sections
with plain-language explanations, worked examples for quantitative subjects, and
embedded check-in questions you have to answer before moving forward. Full LaTeX
rendering throughout for math, chemistry, physics, and economics.
Stage B — Practice
For retrieval and reinforcement. AI-generated questions built from your actual
source material — novel every session, never repeating. Multiple question types
(multiple choice, short answer, worked problems, synthesis) with difficulty
adjusting to your recent performance. If you keep missing the same concept,
Veridian surfaces a targeted review before continuing.
Stage C — Mastery
For verification. Free recall, Feynman-style prompts, and synthesis questions
with no hints or scaffolding. You either know it or you don't.
When you start a new Journey, Veridian runs a short adaptive diagnostic across all modules before placing you anywhere. Each module gets an individual stage placement and an initial FSRS stability estimate. If you already understand half the material, you skip Stage A on those modules and start in Stage B. The diagnostic also seeds your Knowledge Map — the personal mastery record that grows with every session.
Veridian uses FSRS (Free Spaced Repetition Scheduler) — the most accurate spaced repetition algorithm available, significantly more precise than the SM-2 algorithm Anki uses. FSRS models memory as a function of stability and retrievability and computes optimal review intervals for long-term retention. In Veridian, FSRS is additionally calibrated to your exam deadline — backfilling from that date so mastery arrives at the right moment, not six months later.
Available at any time regardless of FSRS schedule. Surfaces the highest-risk material — lowest retention, least time until the exam — and runs an intensive session against it. Designed for the 48 hours before an exam when normal review intervals no longer apply.
Students can publish a completed Journey so others can clone it, plug in their own deadline, and use it as their own. You share the structure — personal learning data never transfers. The library grows as students share what worked.
Every session feeds a persistent Learning Profile: a global Knowledge Map with mastery state per module, a Subject Breakdown with trends per subject, a full Study History log, and an FSRS Health Summary that surfaces plain signals — stable, needs attention, or at risk. All profile data is private regardless of what you publish.
| Layer | Technology |
|---|---|
| Frontend | React, Vite, Tailwind CSS |
| State | Zustand, TanStack Query |
| Backend | Base44 |
| AI | Google Gemma |
| Scheduling | FSRS |
| Math Rendering | KaTeX |
| Privacy Mode | WebGPU (on-device inference) |
git clone https://github.com/OrangJaguar/Veridian.git
cd Veridian
npm install
cp .env.example .env
# Add your API keys to .env
npm run devRequired environment variables are documented in .env.example.
Every feature in Veridian is available to every student, permanently. No premium tier, no Pro plan, no paywall. The students who need this most are often the ones who can least afford another subscription.
This project is licensed under the Business Source License 1.1 (BUSL-1.1).
You may view, fork, and run this code for personal and non-commercial use only. Commercial use, hosting as a competing service, or redistribution without permission is not permitted under this license.
The license automatically converts to MIT on January 1, 2028.
See LICENSE for the full terms.
Built by Sanskar Gupta