Founders School · Powered by Alpha
You write every word — facts, insights, spiky positions. An AI that never writes for you grades them against your evidence and against reality.
Powered by Alpha
Alpha School proved AI can compress academics and lift outcomes. Expertise applies the same rigor to the one thing that compounds: knowing something deeply enough to have a defensible view.
How It Works
The Grader
Every sentence is yours. The model checks it against your own approved evidence, then against reality.
01
Grounded
Does the position actually follow from the evidence it cites? Uncited leaps get caught.
02
Correct
Is the claim actually true? A perfectly-cited falsehood still fails.
03
Spiky
Would an informed person disagree? Consensus restatements score low. Sharp, defensible stances win.
Why It Works
A week after studying, learners who had to produce what they knew kept what rereaders lost. That is why you write every note, insight, and position in your own words here — and the grader checks them, not your reading list.
Recall after one week, Roediger & Karpicke 2006
Solving new problems, interleaved vs blocked practice, Rohrer & Taylor 2007
Live Demo
The real workspace UI, clickable. Switch views in the left sidebar; press Write for the compose flow.
What It's For
Audience
Every grade-A Position is publishable content waiting to happen — a thread, an essay, a podcast talking point. The app surfaces them; you take them into the world in your own voice.
Business
Validated, spiky Positions are the asymmetric knowledge that produces non-obvious ideas and defensible plans. Twenty grade-A Positions beat starting from zero.
You
The bar: pause an expert interview after the question, before the answer, and give an answer at least as informed as the expert's. There is no “done” state — as long as you're learning, the graph grows.
Questions
Never. The hardest deliberate constraint in the product: the AI grades and surfaces. It does not extract facts, draft summaries, or write positions. Every note in your graph is a passage you highlighted yourself, and every graded word is yours.
Three things: that your position actually follows from the evidence you cited, that the claim is true — a perfectly-cited falsehood still fails — and that an informed person could disagree with it. Fall short on grounding or truth and the position can never reach publishable, no matter how polished it reads.
Positions only. Your Score is the sum of your Position grades; facts, summaries, and insights are scaffolding. Sharper, better-grounded positions raise it, and nothing else does.
Each Brainlift is a sealed workspace: one knowledge graph, one source library, one scratchpad, one Score. Six views — graph, outline, tables, sources, scratchpad, publish — render the same graph from different angles; switching views never changes what's there. No tags, no folders: citations are the structure, search and pivots are how you cut across it.
Expertise is a Founders School product, built on the Alpha School model of AI-accelerated, mastery-based learning. Your publishable positions flow to the other Founders School tools, grounding idea grading and audience content.