In the app
Search
What Slingdrop search can find inside your files, and asking your library questions with ⌘K.
Press ⌘K (or Ctrl K) anywhere in the dashboard to search your library. Search looks at titles and file names, and also inside your files: the words in a PDF, the text of a slide deck, what is said in a recording, or what is in a photo. It only ever searches drops you own.
What's searchable#
Files are read once, when they finish uploading. A new version is read again, and search always looks at the current version.
| What | Found by | Plans |
|---|---|---|
| Titles, file names and AI tags | Every keystroke, instantly | All signed-in plans |
| Text files and code | Words inside .txt, .md, .csv, JSON, YAML, logs, HTML and code files (up to 2 MB each) | All signed-in plans |
| PDFs | The text layer of every page, kept page by page | All signed-in plans |
| Office and iWork documents | Word, PowerPoint and Excel files (doc, docx, ppt, pptx, xls, xlsx), OpenDocument, RTF, and Pages, Keynote and Numbers | All signed-in plans |
| Images | A short description of the picture plus any visible text (screenshots, receipts, whiteboards, signs) | All signed-in plans |
| Scanned PDF pages | Pages with no text layer are read like images, and the text lands on the right page | All signed-in plans |
| Videos | A few frames are described in a sentence or two each (what's on screen) | All signed-in plans |
| Audio and video transcripts | Everything that's said, transcribed automatically | Pro and Team |
Matches come two ways: exact words (with prefix matching, so contr finds contract) and meaning, so car insurance can find a policy that only says vehicle coverage. Meaning-based results use multilingual embeddings, so a Swedish query can find an English document. Plain search like this never uses AI credits.
What's never searched#
- Vault drops. They are end-to-end encrypted, so Slingdrop can't read them. They are never indexed, never sent to an AI model and never appear in search results.
- Drops that are in Trash, expired or burned.
- Drops other people shared with you. Search covers your own library.
- Anonymous drops (sent without an account).
Reading images, scanned pages, video frames and speech happens on our infrastructure and AI providers; see Subprocessors. If you don't want a file read at all, send it as a Vault drop.
Ask in a sentence#
Type a whole sentence and press Enter, and Slingdrop AI works out what you mean. Results appear as you type; Enter hands the sentence to the AI.
| You type | What happens |
|---|---|
| the contract Anna sent last month | Finds drops, with filters picked from your library: Anna as the sender, last month as the date, documents as the type. |
| pdfs from this week | Filters by type and date. Each filter shows as a chip you can remove. |
| what's the total on the October invoice? | Answers from inside your files, with numbered sources that open the right file and page. |
| make the Q3 deck private | Proposes the action on the drop it found. Nothing changes until you confirm it. |
| open settings | Takes you to a page of the app, such as Settings, Trash, Requests or Hosting. |
- Filters are built from your own data: the kinds of files you have, the people you send to and receive from, your tags and how drops arrived (uploaded, emailed, a file request or an AI agent).
- Actions it can propose: copy a link, open, download, make private or public, and share with someone by email. It never deletes anything from search.
- Each sentence sent with Enter counts as one AI action from your monthly allowance (see Limits). When you run out, search falls back to plain results.
- On a single drop, Ask about this asks a question about just that drop.
From code#
Search is part of the app and isn't in the public API. From code, list drops with GET /drops (optionally by folder) and filter them yourself, or give an AI agent the list_drops and get_drop tools.