HOW WE ACTUALLY TEACH

Teaching isn't magic. It's a set of decisions.

This is the long version of why Felloclass works the way it does. It's the clearest picture we can give of what happens between the moment you type a sentence and the moment a lesson starts playing.

For most of recorded history, the thing that separated the kid who "got it" from the kid who didn't was almost never raw intelligence. It was access. Access to a teacher who noticed exactly where that kid was stuck, and rebuilt the explanation until it landed. Information has been cheap for two hundred years. Attention from a teacher has always been scarce. Felloclass is a bet that a well-designed AI teacher can finally give every learner that attention.

  1. One learner at a time, always.

    Every course is generated for a single reader. There is no shared syllabus. There is no "other students also asked." The teacher is in a room with you.

  2. Start from what you know, not chapter one.

    Before the first lesson runs, Felloclass calibrates, briefly and conversationally, so we don't waste the first forty minutes re-explaining what you already have.

  3. The blackboard, not the video.

    Pre-rendered video freezes a teacher's decisions. A live blackboard re-draws itself based on what you just said. We build every lesson to happen now, not to play back.

  4. Honesty about what we don't know.

    When Felloclass isn't sure of a claim, it says so, and shows its sources. When a learner asks a question the teacher can't answer well, it logs it, and we fix the lesson.

What we measure

  • Did the learner keep going? Not streaks. Actual returning sessions.
  • Did the learner outgrow the next lesson? Lessons that are too easy or too hard both fail.
  • Did the learner answer a question at the end that they couldn't answer at the start? The simplest possible test of whether anything was taught.

What we refuse to do

  • Put the game in front of the lesson. Streaks and XP exist, but they sit quietly around the work. The lesson is the thing.
  • Rank learners against each other.
  • Show "completion rates." The only completion rate that matters is yours.
  • Promise outcomes we can't prove.

I'll change this page whenever I change my mind. The date of the last edit is at the bottom.

Last updated: 2026-04-19