But I also didn't want an app that forced me to do all my reading on screen - I want to read physical books on the couch, listen to a history podcast when I walk, watch a documentary with my wife. And still track what I learnt, and maintain it...

So I built an app that let's me track which books I am reading by taking a picture of the cover (it automatically researches them, inserts TOC and content, autosyncs with Kindle progress etc), and generates review questions based on book content and generated "curricula".

But to map the things I already know, I designed a "knowledge elicitation" using voice - it gives me some prompts (what do you know about Plato), I talk into the mic while walking, and it maps what I know, my mistakes, my gaps, and also what I'm curious about. What I'm curious about (how was Plato different from Socrates?) becomes a micro-learning card that is scheduled together with the review cards...

So far it's proving really valuable. Lot's to tune as I keep using it, but I feel the combination of deep learning offline, and review digitally when you have a few minutes free, is the best of both worlds.

Repo here: https://github.com/houshuang/petrarca (point Claude Code at it, requires a VM if you want it on your phone, some API keys etc)