How we built a 3-minute personalized podcast.
The technical decisions, the model choices, and what we threw away to keep MemoCast under three minutes per episode.
Capture is solved. Recall is the new bottleneck. Here's why we built MyMemo around the second half of the problem — and why “where did I read that?” is the most expensive question in knowledge work.
Read article →The technical decisions, the model choices, and what we threw away to keep MemoCast under three minutes per episode.
Why every answer in MyMemo links back to a source. The trust model behind a personal AI you'd actually want to use.
The folder is the enemy. A short essay on why search has finally caught up with PKM, and what that frees you to do.
Inspired by Tim Urban's piece, on how MyMemo tries to become a chef-grade tool without scaring off the cooks.
Inside the heuristic that routes your queries to GPT, Gemini, or Claude — and the cost trade-offs we hide from you on purpose.
Five small habits that turn your saved articles into something you can build on a year from now.