How Teachback is different
The "explain it to an AI" pattern isn't new there's a whole crowded field of Feynman-technique study apps doing some version of explain → get feedback. Most of what's out there, though, shares the same gaps:
- They grade you on a rubric. Teachback grades you on your own words. Every "clear point" and every "gap" in your report is grounded in an exact quote pulled from your transcript — not a generic "clarity score," but the actual sentence that proved (or didn't prove) you understood it.
- They evaluate one big explanation. Teachback has a real conversation. The AI student reacts turn-by-turn to what you just said including catching it when you contradict something you said three messages earlier, which requires genuine conversational memory, not a single-shot evaluation.
- They bury the technique in a feature list. Teachback does one thing. No flashcards, no quizzes, no subscription tiers just the loop that matters: explain it, get caught, see exactly what to fix.
- They ask for a signup before you can try it. Teachback doesn't. No account, no paywall — straight from the landing page into the conversation.
None of this makes the underlying idea original the Feynman technique is decades old, and "teach an AI" is an established pattern. What we focused on instead was making the one loop that actually matters (explain → get genuinely questioned → see the receipts) as tight and honest as possible, rather than surrounding it with features that dilute it.
Log in or sign up for Devpost to join the conversation.