Bloom’s 2-sigma problem · 1984
A tutor’s method,for everyone.
Bloom is an AI tutor. Give it a topic or your own materials and it writes a syllabus, then teaches one lesson at a time, writing each next lesson from what you highlighted, what you asked and what you wrote back.
2000 free credits on sign-up, no card needed
01 Why it works
One-on-one tutoring works. It just doesn’t scale.
In 1984 the educational psychologist Benjamin Bloom reported that students taught one-on-one performed about two standard deviations better than students in a conventional classroom, which puts the average tutored student at roughly the 98th percentile of the classroom group.
He called it the 2 Sigma Problem: the effect is real, but one-on-one tutoring is too costly to give every student.
What Bloom tries to do is hand you the way a good tutor teaches.
+2σ Bloom’s 1984 report: tutored students about two standard deviations above the classroom group
A finding about human tutors, not a promise about Bloom.
A tutor sees where you are
A good tutor notices what you got and what you didn’t, then changes the next explanation. Bloom works from the same signals: what you highlighted, what you asked and what you wrote back.
You move on when you’re ready
Your syllabus is a checklist of things to master, and each lesson builds on the last one and on your answers. When the list is done, Bloom writes an assessment and a course summary.
Inspired by the finding, not a promise of it
The 2-sigma result came from studies of human tutors. Bloom borrows the ideas behind it (attention, adaptation, mastery) but does not promise you the same result.
02 How it works
One loop, repeated until you’ve got it
The rhythm a good tutor follows. The difference is that each next step is written from what you actually did.
Syllabus
Name a topic or bring your own materials. Bloom writes a syllabus of things to master first.
Lesson
One focused lesson at a time, instead of a wall of text.
Highlight and ask
Select the sentence that lost you and ask about it. The answer appears beside the passage, and you can keep following up until it clicks.
Write feedback
Answer the reflection questions: what was clear, what wasn’t, and what you want to explore next.
The next lesson is written from you.
Bloom writes it from your answers, highlights and feedback, until the syllabus is done. Then you take an assessment and read a course summary.
03 Three ways in
Start from whatever you have in front of you
No setup. Choose one when you create a course.
A topic
Billed by tokensGenerate an outline and lessons from a topic, with optional notes or files. Billed by actual token usage.
Your own materials
Billed by tokensLearn from your PDF, text or Markdown source. Generation and questions are billed by actual token usage.
Project files
Free importImport files or folders as readable pages for free. AI questions are billed by tokens.
04 Features
Built around how tutoring works
Each part of Bloom is there to keep the lesson matched to you.
A syllabus with a point
Your goal becomes a syllabus of concrete things to master. Choose a simple, standard or in-depth version when you create the course.
Ask on the passage
Highlight any passage in a lesson or a source document and ask about it. Answers stream in as they are written, and follow-ups stay in the same thread.
Lessons that remember
Each new lesson reviews your reflection answers, replies to the passages you highlighted, and then moves on to what is still unclear.
Your sources, cited
For PDFs and Markdown files, page numbers and sections are kept, so lessons can point back to where something came from.
What to learn next
Bloom suggests three topics that build on what you have learned. Save ideas for later or start one straight away.
Your learning profile
A calendar and activity map show what you studied each day, along with your streak and how many lessons you have read and questions you have asked.
05 Principles
Four ideas doing the heavy lifting
Mastery learning
The syllabus is a checklist of things to master, each one something you can explain and use. What isn’t mastered yet keeps coming back in later lessons.
Socratic questioning
Questions first, answers second. Every lesson ends with reflection questions, and your answers carry into the next one.
Spaced review
Each lesson opens by revisiting the last one’s questions, so what you’ve learned gets another pass inside the new material.
Adaptive path
The goal is fixed, the route isn’t. The next lesson is written from your highlights, questions and feedback, not from a fixed lesson plan.
06 Pricing
Buy when you need it, pay as you use it
Credits come in one-time packs: no subscription and no auto-renewal.
Usage based
Peak model token rates × 3, with separate input, output and cache prices.
1 credit = CNY 0.01
Fractional credits; no rounding each call up to a whole credit.
Transparent ledger
View purchases, bonuses, charges and refunds.
Three credit packs
- Starter
3,577 credits · valid for 12 months
$4.90CN¥35.77 - StandardRecommended
14,527 credits · valid for 12 months
$19.90CN¥145.27 - Pro
29,127 credits · valid for 12 months
$39.90CN¥291.27
Credits you buy are valid for 12 months from payment. The 2,000 free credits new accounts receive stay valid for 365 days from sign-up.
How credits are calculated
Pay for actual token usage
1 credit = CNY 0.01. Input, output and cached tokens use the model’s peak rates, multiplied by 3. Calls within one action are combined; internal repairs are paid by Bloom.
Initially reserves up to 100 credits, adding more from your balance if needed. Actual usage is settled on completion. USD rates convert at CNY 7.3/USD.
07 FAQ
Questions people ask before starting
Straight answers about how Bloom works.
What exactly is Bloom?
How is that different from chatting with an AI assistant?
Will Bloom get me the 2-sigma result?
What can I upload?
How do credits work?
What happens to my materials and courses?
Pick something you’ve been meaning to learn
Sign in, name a topic or drop in a file, and Bloom writes your first lesson.