How do I find a bookmark I forgot the title of?

Describe what you remember and let semantic search do the rest. In Marqly, type something like “that video about sourdough starters” or “article on pricing psychology from a founder blog” — the search matches meaning, not exact words, across titles, page content, your highlights, and video transcripts. You don’t need the title, the site name, or whatever tag you would have filed it under.

Why keyword search fails at this

Browser bookmark search and history search match literal text in titles and URLs. But the piece you’re trying to find rarely cooperates: the title was clickbait that never mentioned the actual topic, or the URL is a string of numbers, or you saved it three months ago and all that’s left in your head is the idea. When your memory holds a concept and the search box demands a keyword, you end up scrolling through hundreds of entries by hand.

What semantic search actually matches

Semantic search compares the meaning of your query against the meaning of everything you’ve saved — not just titles, but the content of saved pages, the passages you highlighted, and the transcripts of saved videos. That last part matters more than it sounds: a video’s title almost never contains the detail you remember from minute 23, but its transcript does. The mechanics — and what semantic matching does differently from a keyword engine — are laid out in what is semantic search and how to search bookmarks with AI.

Writing a query that works

Say what the thing was about, the way you’d describe it to a friend: topic, angle, any specific detail that stuck (“interview where the founder talked about firing himself”). Fragments are fine. If the first phrasing misses, describe a different aspect of the same save — a different door into the same room often opens it. A longer walkthrough with worked examples: how to find a saved article when you’ve forgotten the title.

When one search isn’t enough

If a single query doesn’t land it, Ask is the conversational version of the same hunt: Marqly’s Pro assistant over your entire library, opened with ⌘J (Ctrl+J on Windows). It’s a multi-turn chat — you keep describing the thing from a new angle until it surfaces — and its answers come with numbered citations back to the saves they were drawn from, so you can tell a real match from a plausible one.

The honest limits

Semantic search and Ask are Pro features — see what’s in Marqly free vs Pro. On the free plan, keyword search still covers your whole library, so the same hunt works when you can supply any exact word: a site name, a phrase, a tag you remember using. And one boundary no tier removes: search only covers what you actually saved — the article you read on your phone without bookmarking it isn’t in any index.

The prevention side

Retrieval is half the battle; the other half happens at save time, automatically. Marqly auto-tags every save with AI (Pro), so even a library you never manually organized stays searchable by topic as well as by meaning.