Memoir: AI Study Assistant

Memoir AI turns textbook photos into AI-powered lessons, flashcards, and quizzes. Multi-profile, gamified, offline, and multilingual—learn faster anywhere.

Comment

Inspiration

Memoir AI story is not very original. I came across a similar app that did part of the job, but the path from “I just snapped a page” to “I’m actually learning” felt too slow and fragmented. I wanted an experience that turns photos into understanding in one flow, adds gentle motivation, gamification, and works well for families too. Memoir is that attempt: practical, fast, and intentionally delightful.

What it does

You take photos of your textbook or notes and Memoir converts them into a clean lesson: a concise summary you can read or listen to, adaptive flashcards to reinforce key ideas, and a quiz to check understanding. It keeps you engaged with XP, streaks, and achievements, and it lets families switch profiles so each learner has their own progress. When you’re ready to go further, Premium unlocks unlimited lessons and advanced features without complicating the experience.

How we built it

The app is built with Ionic Vue and Capacitor for a smooth cross‑platform shell, TypeScript for safety, and Vite for fast iteration. The AI pipeline accepts photos, calls a backend to generate the lesson, the flashcards, and the quiz, and uses a retry with backoff to survive mobile network hiccups. For audio, I added text‑to‑speech playback using ElevenLabs, plus on‑device caching so summaries remain available offline. Subscriptions run through RevenueCat’s Capacitor SDK: configure on launch, fetch offerings, purchase or restore, and gate features based on entitlements across the UI. The app is ~60% vive-coded, thanks to Claude Code and Cursor.

Challenges we ran into

Two product challenges shaped the app. The first was making the AI output truly pertinent to the material you capture—not a generic summary. I worked on keeping context faithful to the photos, passing just enough information about the learner’s profile and goals, and steering the result toward concise, factual explanations with vocabulary that matches the level. Multi‑photo captures are treated as one lesson so the flashcards and the quiz reflect the whole chapter, not just a single page.

The second was retention. I leaned into gentle gamification—XP, streaks, small achievement moments—and paired it with local notifications. The idea is to nudge you back at the right time to continue a lesson or review cards without feeling spammy. That loop—create, review, quiz, see progress—became the backbone for turning “I tried it once” into a steady study habit.

Accomplishments that we're proud of

What makes me proud is how little effort it takes to go from paper to practice. You snap a couple of pictures, and within moments you’re looking at a tight summary you can read or tap to hear, then you’re flipping through smart cards and jumping into a quiz—no copy‑paste, no configuration, no friction. Multi‑photo capture becomes a single coherent lesson so it feels like studying a chapter, not a page at a time. The flow is intentionally calm and direct, with subtle feedback and small celebrations that make progress feel rewarding rather than like homework.

What we learned

Deep AI integration is more than “call an API.” For lesson, card, and quiz generation, the hard work is in shaping consistent inputs from messy photos, protecting context, and guiding the model toward concise, level‑appropriate output. On the audio side, working with ElevenLabs taught me practicalities like chunking long text without audible seams, caching voice output on‑device, and falling back gracefully when a request fails. For gamification, the lesson was about rhythm: small, frequent wins; streaks that recover quickly; gentle haptics and micro‑animations; and local notifications scheduled to meet the learner right when review is most effective. When the AI and the game loop support each other, studying becomes a habit.

What's next for Memoir: AI Study Assistant

Two directions excite me most. First, “talk with your lesson”: a conversational layer where you can ask follow‑ups about the material you just captured. Second, deeper learning loops—more gamified moments and true spaced repetition for cards and quizzes. I’ll also expand languages and accessibility, keep trimming offline bundles, and use RevenueCat Offerings to iterate on family plans without shipping app updates.

Built With

Share this project:

Updates