Inspiration

Most students confuse familiarity with understanding. You watch a lecture or read book, nod along, feel like you "get it" — then blank out on the exam. The real problem isn't lack of resources. It's that passive consumption feels like learning but isn't. We wanted to build something that forces real understanding, not just exposure.

What it does

ExplainBack is built on the Feynman Technique — if you can't explain something simply, you don't truly understand it. Type any topic you just studied, explain it back to AI, and get structured feedback on exactly what you understood and what you only thought you did.

After explaining you get in response,

  • What you got right — on basis of what you actually explained
  • Gaps found — concepts you missed
  • Misconceptions — what you got wrong, corrected simply
  • Understanding score out of 10
  • Visual Concept Map showing subtopics as Understood / Partial / Missed
  • AI quiz with option to practice 5 more questions
  • Quick Recap before quiz
  • Dynamic topic suggestions based on your study history

How we built it

Built entirely with MeDo through multi-turn conversations. We described each feature in prompt and iterated on MeDo's output to refine the app — voice input, text-to-speech, dark mode, multilingual support, localStorage history, and the concept map visual were all added through iterative prompting.

Challenges we ran into

Getting the AI to evaluate only what the user actually said — not hallucinate extra points or fabricate examples the user never mentioned. Also making the Concept Map feel visually clear and meaningful rather than just a list.

Accomplishments that we're proud of

The Concept Map of Understanding — a visual tree showing exactly which subtopics were understood vs missed. This single feature transforms ExplainBack from a chatbot into a genuine learning intelligence tool. Also proud of the dynamic topic suggestions that adapt to each user's study history.

What we learned

The Feynman Technique as a product concept is deeply underutilized. Also learned how to build and iterate on a full-stack AI app purely through conversation with MeDo.

What's next for ExplainBack Learning Tool

Spaced repetition reminders, progress analytics across sessions, and a classroom mode where teachers can assign topics for students to explain back.

Built With

  • gemini-2.5-flash
  • medo
  • react
  • speechsynthesis-api
  • supabase
  • typescript
  • web-speech-api
Share this project:

Updates

Submission history