BLOOM'S 2-SIGMA PROBLEM · 1984
Bloom is your personal AI tutor. It designs a syllabus, teaches one lesson at a time, reads your every annotation, and reshapes the next lesson to exactly where you are.
THE PROBLEM
In 1984, educational psychologist Benjamin Bloom found that students who received one-on-one tutoring performed two standard deviations better than those in a conventional classroom — moving the average student into the top 2%.
The catch: human tutors don’t scale. Bloom called it the “2 Sigma Problem.” The effect is real, but you can’t give every student a personal teacher.
Bloom closes the gap with AI.
+2σ
standard deviations above the classroom average
HOW IT WORKS
The same rhythm a great tutor follows — automated, and tuned to you.
AI designs a mastery-based syllabus from your topic and any reference material you paste.
One lesson at a time. Never a wall of content dumped at once.
Select any passage and drop a ??? exactly where you’re confused.
Answer the thought questions and write down what clicked.
AI reviews your answers, addresses every ???, then teaches on.
When every mastery item is checked, an evaluation confirms it.
A personal knowledge map, auto-written from your whole journey.
↺ LOOP Then it repeats — each pass tuned tighter to you, until mastery.
FEATURES
Paste textbook chapters, papers, or notes — the AI shapes the curriculum around them.
Highlight any passage and mark precisely what’s unclear, right where you read it.
Every lesson targets the specific gaps you revealed in the last one.
Jump between lessons freely while you read, without losing your place.
Watch the next lesson write itself, in real time.
Streaks, annotations, mastery — your effort, made visible.
THE SCIENCE
Don’t move on until the concept is truly understood.
Ask questions and guide thinking — don’t hand over answers.
Each lesson opens by reviewing the last. Memory, reinforced.
Fixed goals, flexible route. The path bends to you, not the other way around.
TWO WAYS TO USE
For power users who live in Markdown editors. No backend to run — just Claude Code and a folder.
For visual learners. A polished browser app with text-select annotations, streaming, and progress tracking.
QUICK START
git clone https://github.com/Li-Evan/Bloom.git
cd Bloom
cp .env.example .env # add your LLM_API_KEY
# backend
cd backend && uv sync && uv run uvicorn app.main:app --reload --port 8000
# frontend (new terminal)
cd frontend && npm install && npm run devOpen http://localhost:5173, click New Course, enter a topic, and start learning.
Open source. MIT licensed. Bring your own LLM key.