Bridge

Notes & classification

A note is the smallest unit of capture in Bridge — type, paste, or dictate anything that comes to mind. There's no template, no category, no required title. The Notes tab is the entry point.

What happens when you drop a note

Within about thirty seconds of capture, two passes run on your note.

Auto-classification

Bridge classifies every note as one of two kinds:

  • Expandable — a thought worth elaborating on. Bridge will auto-generate a one-paragraph elaboration (on Discover tier and above).
  • Reference — worth keeping for context, but not worth expanding. Quotes, links, source material, snippets you might pull from later.

You don't pick the classification — Bridge infers it from what you wrote. You can override it in the note detail page if you disagree.

The on-capture scan

Bridge runs a quick scan on every note regardless of tier. It extracts:

  • URLs — surfaced as link chips.
  • Hashtags — anything starting with #.
  • Code blocks — fenced ``` blocks rendered as monospaced.
  • Co-capture timing — notes captured within minutes of each other get a structural connection edge.

These signals populate the graph view at the Free tier with no AI cost.

What you see in the note detail

Tap any note in the Notes tab to open the detail page. The detail shows:

  • Your raw note text, exactly as you wrote it.
  • The classification chip (Expandable or Reference) with an override option.
  • The extracted structural elements (URL chips, hashtag chips, code blocks).
  • The general expansion paragraph (Discover tier and above).
  • Thorough expansion modes — Idea / Feature / Product (Build tier and above).
  • A Star to pin the note to the top of your Notes tab.
  • An Archive to remove it from the main view without deleting.
  • A Start a venture button that pre-seeds the venture picker with this note's constellation (Build tier and above).

What Bridge does not do

  • No editing of captured notes — once captured, the raw text is the raw text. (We're researching cascade re-processing for edits.) The expansion and the on-capture scan can re-run, but the source text is immutable.
  • No required tagging — hashtags are optional. The graph forms from semantic similarity, not from your manual organizing.
  • No folders — there's no folder hierarchy. Search, classification chips, star, and archive are the organizing affordances.

Next