Veridian — Devlog 01

Sophomore year I walked into AP Chemistry, AP US History, and AP Psychology on the same schedule and realized for the first time that I had no idea how to actually study. Not in a “I’m lazy” way — I’d just never had to before. A quick review the night before a test was always enough. Nobody told me that was going to stop working until it already had.

I tried everything. Re-reading notes felt like going through the motions. YouTube videos felt productive right up until I closed the tab and remembered nothing. Nothing told me whether what I was doing was actually working until the test came back with a number on it, and by then it was too late to change anything.

I looked for something better. What I found were tools built for students who already knew how to study. None of them knew my material, my exam date, or which concepts I genuinely didn’t understand versus which ones just needed one more pass. So eventually I just built the thing myself.

That’s Veridian.

What It Does

You upload your notes, a textbook chapter, a study guide — whatever you actually have for your class. Veridian pulls out the knowledge structure inside that material and builds you a Journey: a personalized study plan that figures out what you don’t know, drills you on it, and tells you exactly where you stand before the exam. Not after.

The whole system is driven by FSRS, the most accurate spaced repetition algorithm out there, calibrated to your actual exam deadline so your review timing is mathematically optimized. The AI runs on Gemini Flash. And it’s completely free, permanently. Not a launch promotion. The students who need this the most are usually the ones who can’t afford another subscription.

How It’s Built

This is the part I’ve spent the most time on. Here’s the core architecture:

Journeys are the main container. Inside each Journey are Modules, which are individual concept units extracted from your source material. Every Module moves through three stages:

Stage A — Learn: A guided Learning Guide walks you through the concept with plain-language explanations, worked examples, and embedded check-in questions after each section. It’s not passive reading — you have to answer something before you move on.
Stage B — Practice: AI-generated questions that never repeat. Every session builds new questions from your actual source material, so you’re always being tested on understanding, not card recognition.
Stage C — Mastery: Active recall with no scaffolding. This is where you prove you actually know it.

Underneath all of it is the FSRS engine deciding exactly when each concept needs to surface again based on your performance history. It’s not random — it’s a mathematical model of how human memory actually works.

The Home screen is built around a single question: what should I do right now? Every Journey with something due today shows up as an action card with the activity type, a time estimate, and a single launch button. You open the app and start. That’s it.

There’s also a Community Library where students can publish their Journey so others can clone it, plug in their own deadline, and use it. You share the structure — your personal progress and data always stays yours.

Tech

React + Vite + Tailwind + Zustand + TanStack Query + Base44 for the backend. Gemini Flash for AI. FSRS for scheduling. Full LaTeX rendering throughout for math-heavy subjects.

Where It’s At

Veridian is currently closed-source. You can sign up for the live public version at veridian.study.

Free forever. No paywall, no AI upsell, no premium tier. More devlogs coming as I keep iterating.

— Sanskar