How I finally got to use Dataview in Obsidian (with AI/File Organizer 2000)

Dataview in Obsidian turns scattered note metadata into instantly readable tables—so users can scan, compare, and filter information without opening every note. The core payoff is speed: once notes carry consistent fields (like author, genre, ratings, or page counts), Dataview can generate a live table view that updates as the vault changes. The catch is that building and maintaining those tables takes time, and writing the required Dataview syntax (plus manually filling many note properties) can become a bottleneck.

A practical workflow in the transcript pairs Dataview with AI-driven automation to remove that bottleneck. First, the user demonstrates a Dataview table for a “books” collection stored in a specific folder. The table pulls properties from each note—such as the book title, author, genre, and a personal rating—and renders them in a single grid. The table is generated using Dataview’s query syntax, including targeting notes by folder and selecting which properties to display.

Where the workflow shifts is in eliminating repetitive data entry. Instead of typing every property for each book note, the user uses AI templates (via “falaz 2000 AI chat,” though the approach is described as compatible with other chat tools like ChatGPT). A “book template” prompt instructs the AI to detect when an uploaded item looks like a book and then extract or generate a structured set of fields: title, author, genre, short summary, page number, year, Goodreads rating and URL, Amazon URL, personal rating, and notes. This template is applied automatically when new files arrive.

To test it, the user deletes the existing book entries, then takes photos of four book covers and sends them into Obsidian using an “Add to Obsidian” shortcut that routes files into an inbox processed by a file organizer. As the images are processed, the AI recognizes them as books and populates the template fields. The results include not only basic metadata like author and genre, but also page counts and external links (Goodreads and Amazon). The user then manually adds only the fields that the AI prompt left blank—such as personal star ratings.

With the enriched properties in place, Dataview becomes more powerful because it can filter and re-render the same table based on conditions. The transcript shows examples of filtering by Goodreads rating (e.g., only books above a threshold) and by page count (e.g., maximum pages). The same pattern is positioned as reusable beyond books: the user cites prior examples involving receipts and calorie tracking, where photos are converted into structured entries that feed Dataview tables.

Overall, the central insight is that Dataview’s value scales when AI handles the tedious part—turning unstructured inputs (like photos) into consistent note properties—so tables and filters remain accurate with minimal manual work.