5 updates to Logseq and my workflows | Personal Knowledge Management

Logseq’s most consequential recent upgrade is the ability to search inside PDFs—turning stored documents into something that behaves like fully searchable knowledge, not just a static attachment. With the familiar Ctrl+F workflow, searches now scan within PDFs and jump to matching instances, including multiple hits across the same page. While the interface can feel a bit clunky (results appear across both sides of the viewing area), the practical payoff is immediate: finding a specific term inside a PDF becomes as fast as searching a text page.

The other major shift is how Logseq handles structured metadata and references. Properties have been streamlined: instead of relying on the older org-mode-style “properties” container approach, users can now create key/value properties more directly using “::” and select from previously used keys. The key idea is context: a link like “Dario” becomes far more meaningful when paired with a property such as “birthday,” and the same pattern can apply to article metadata (producer, tags, and other attributes). This matters because it improves both organization and retrieval—especially as the database grows (the creator cites having hundreds of input properties).

Filtering and reference navigation also get sharper. Link references in filters are now ordered by the number of links, so high-signal pages surface first instead of drowning in an unordered list. Inline block references are another usability win: rather than jumping away to see where a block is used, Logseq can show references in place, preserving context while still letting users trace usage. For someone building a reading-and-writing workflow—like processing quotes from a book into scratch pads—this reduces the “lose your place” problem.

Finally, namespace queries and the broader workflow philosophy behind them tie the updates together. Namespace queries let users pull up hierarchical structures—anywhere in the database—by typing namespace and a namespace term (e.g., BJJ positions, intentions, or scratch pad areas). That capability pairs with the creator’s emphasis on self-updating indexes: events, learning materials, and organizational hierarchies can be surfaced quickly without manually maintaining separate tables of contents.

On the personal workflow side, the creator adapts to these changes by tightening the input/output distinction using properties and by visually separating borrowed text from original writing with Markdown quote formatting (quotes for “not mine,” plain text for “mine”). They also move away from an overly rigid Zettelkasten-style approach that required turning every thought into an atomic note; instead, they keep content in blocks when it’s still “in process,” using tags, headings, and MOCs-like pages as flexible containers. To make the system more sustainable, they add custom CSS to reduce visual friction (e.g., styled tag buttons and inbox highlights) and rely on native outline behavior—Markdown headings that can be expanded and collapsed—to manage dense pages. The overall message is pragmatic: Logseq’s improvements make it easier to search, filter, and navigate; the workflow changes focus on returning to information quickly without forcing perfect structure too early.