Learnify

An agentic tutor that teaches through generative UI.

Learnify turns any topic into a live, adaptive tutoring session. Instead of serving pre-written courses, the tutor — Claude on AWS Bedrock — generates each screen as you learn: explanations, checks, flashcards, diagrams, and exercises stream in as declarative UI and adapt in real time to how you answer.

What it does

• Generative classroom — The tutor streams each screen live as declarative JSON (A2UI). Explanations, checks, flashcards, diagrams, ordering and matching exercises, and code snippets render from a vetted widget catalog and adapt in real time to how you answer.
• Mastery engine — Every topic decomposes into a prerequisite concept graph. Per-concept mastery is tracked with spaced repetition (EMA + SM-2), and a daily Today dashboard surfaces what's due, what's weak, and your streak — turning learning into a habit.
• Tutor memory — The tutor remembers how you learn across sessions — your style, interests, and past misconceptions — so a football fan keeps getting football analogies, even on an unrelated topic.
• Learn from anything — Attach a PDF, article URL, or YouTube video and the tutor teaches grounded in your sources, with citation chips that jump to the exact quote, page, or timestamp.
• Zero-friction demo — Run a full live session with no login. Sign in to persist your mastery graph and memory across devices.

How it works

A session is an agent loop: Claude emits UI through strict tool calls as A2UI blocks, the client renders them from a sandboxed widget catalog, and your interactions (answers, recall ratings, clicks) stream back as the next turn. Objective answers are graded server-side and feed the mastery engine; attached sources are passed as citation-enabled context. The renderer only ever interprets data — it never executes model-authored code.

Stack

Next.js 15 (App Router) · TypeScript · Tailwind CSS v4 · @anthropic-ai/bedrock-sdk (strict tool-use A2UI emission) · A2UI declarative UI · Neon Postgres + Drizzle ORM · Supermemory · Vercel

Getting Started

Prerequisites

• Node.js 18.18+ (20+ recommended) and npm
• A Neon Postgres database (required at runtime)
• AWS credentials with Bedrock access to Claude models (required at runtime)

Setup

1. Clone and install:

   git clone https://github.com/ahmedfahim21/Learnify.git
   cd Learnify
   npm install

2. Copy the environment template and fill in your keys:

   cp .env.example .env

   See .env.example for the full list (AWS Bedrock credentials, model IDs, and DATABASE_URL).

3. Apply the database migrations to your Neon database:

   npm run db:migrate

4. Start the dev server and open http://localhost:3000:

   npm run dev

Scripts

Script	Description
npm run dev	Start the Next.js dev server
npm run build	Production build
npm run lint	ESLint
npm run db:generate	Generate a Drizzle migration from the schema
npm run db:migrate	Apply migrations to the database
npm run db:studio	Open Drizzle Studio

Legacy

The 2023 codebase is archived under legacy/ with full git history preserved. Follow the rebuild in issue #50.