Blog
-
March 2026
Seven iterations of testing Lectr’s recommendation prompt, and how structured prompting closed the gap between raw and anonymised profiles.
-
March 2026
How Lectr’s Recommendations Work
How Lectr generates book recommendations without sending your book titles, authors, notes, quotes, or tag names to a server.
-
March 2026
Thematic search lets you search your notes and highlights by concept, not just keywords.
-
March 2026
Replacing a Bloom filter with a simpler client-side approach for deduplicating recommendations.
-
March 2026
Testing Lectr’s Recommendations
I tested Lectr’s recommendation feature against my own library of 521 books. Here’s what the experiments showed.
-
March 2026
A short tour of how I use Lectr.
-
March 2026
I laid out every version of my reading app side by side for the first time. What irritation teaches you that theory can't.
-
March 2026
Notes on taste, restraint, and the decisions that matter when your AI coding tool can build anything you ask for.
-
March 2026
I wanted a reading notebook that was private, fast, and didn't try to do too much. Nothing existed, so I built one.