Highlights & labelling

Highlights are how raw documents become usable evidence: a highlighted passage with labels is citable, searchable, and countable.

Creating highlights

Open an uploaded document and select a passage to create a highlight. Each highlight records the exact excerpt and its position in the source, so it always links back to context. When a research item has highlights, a Highlights chip appears on its list row.

Labelling

Apply labels to highlights to build a coding scheme — pain points, needs, feature requests, quotes, or your own taxonomy. Labels come from a shared label library, so the same scheme applies to every source in the project.

• Use Filter by label in the research list header to narrow the list.
• Open Labels… from a list row menu, or Labels on a detail view, to edit label links for notes and research items.

AI-proposed highlights

Crux can read a document and propose highlights with suggested labels, using the Analyse & Label skill in the Agent. Tell it what to look for and it comes back with passages, labels, an Analysis frame of questions, assumptions, evidence, and gaps, and the links between them. The output is always a proposal: review each suggestion, edit the excerpt and the labels, keep the good ones, and dismiss the rest. See Reviewing AI output for how review works across Crux.

Using highlights downstream

• Search — highlights appear in project search (⌘K) with their labels, and you can browse by label from an empty query.
• Ask — labelled highlights are prime citation material when Ask answers from your research graph.
• Evidence links — experience entities can cite highlights directly, connecting a design decision to the exact quote behind it.

Good practice

Keep the label list short at first (5–10 labels) and split a label only when a theme clearly needs it. A tight codebook makes both human review and AI suggestions more consistent.

Next

• Analyse & Label
• Search across your project
• The research graph